The original war · archived
August 1–31, 2026 · 43 players · 18,944 scored posts
The first war's roster, follower records, tweets, and results are preserved separately from round two. Division champions are ranked by days won, then total points, using the original rules. The original site had no end-date filter, so all its stored eligible posts remain counted here.
Back to Twitter Wars 2 →🏆 Champion: @venividinotavi
🏆 Champion: @Liface
🏆 Champion: @funplings
🏆 Champion: @__drewface
🏆 Champion: @qiaochuyuan
1311 posts · 1,987,000 points
72,300 points
Look, I don't want to be that guy, but at what point do we consider just wiring each of these 278 students' families a $52,349 check for non-taxable spending on their child's education or investment accounts. Like... maybe as an experiment for just ONE school. What if we try? https://t.co/e5OYTZJp8d
61,200 points
just in case you're curious for a benchmark -- my wife and I chose to have a kid while living in NYC in market rate housing on ~$100k/yr; with a caveat that we have both been software engineers, and we had a lot of confidence in our ability to figure things out. https://t.co/or5sImNa9R
55,200 points
I just don't understand why we can't figure this out. We need universal basic dormitories, at the very least. Have you ever been in one of those storage facilities? My $100/mo bushwick storage unit was nicer, with better A/C and facilities, than my $1400/mo bushwick bedroom https://t.co/bda25YSS6Q
40,400 points
@christineist it's even weirder from the first person. Sometimes it feels like I'm on another side of a wall of glass. I can see the society that I used to be a part of, but I can't quite participate in it, certainly not the way I used to. And I wonder, is this how all parents felt?
40,100 points
Luckily, my wife and I are both very easygoing and love each other very much. However, I have never struggled to be a good partner as much as the toddler era, and it's really not close. I suspect a big part of a long-term, healthy marriage is optimizing for surviving a toddler. https://t.co/u6x2voWV14
33,800 points
> be 15-17 > be internet-native > want to meet up with friends from internet > can't, you're in jail ("school") > want to start a business > can't, jail > want to learn real life lessons > can't, jail > ask 30-35 yo for advice > "you are illiterate" > no one understands > depress https://t.co/pF5133zuGM
24,400 points
Hear me out: I want more educational spaces like the Climbing Gym. > You walk in > There are almost no explicit instructions > As long as you keep coming back, you get better > It's very obvious how to get better > Progression is clear and obvious > It is social and motivating https://t.co/U6O3jBdK6w
18,900 points
the first time my wife and I had a sleepover I asked "Do you want anything for breakfast?" and she responded "What if we split a Soylent and an RX bar?" and I knew immediately that I was going to marry her. https://t.co/fYNMrHqXmw
16,400 points
In 2016 I fell in love with college 2018 first real job 2019 I met my wife🌹 2020 our first coliving house 🏡👬 👭👫 2021 Fractal 🎨 2022 Marriage 💍 2023 FractalU 📚 2024 👶 🧑🍼 ❤️ + Fractal Tech 2025 daughter turned 1, walking 👣 2026 daughter turned 2, talking 😍 😁 💬 https://t.co/9soZGNdzTj
16,200 points
@MarkovMagnifico fwiw, a concept called "layering" is well studied and is a part of many curriculum design processes. if you were a teacher, how would you motivate your students to use spaced repetition, and how would you design the cards?
16,000 points
>be me >open robinhood after getting a letter saying they are going to send all the money to my state because my account is inactive > i was invested in AMD > 590% return > i only had $250 principal, smh > last sale on the account was $1000 of NVIDIA (2021) > 1800% return missed
14,200 points
ahahahah!!! I finally have a skill that just works to get rid of the overwhelming micro-copy that Claude adds to websites, slide shows, and game UIs!! here's a demo of the before / after on a random "Artemis mission" website I generated: https://t.co/uGJSpwSd7k
13,500 points
i know it's cringe or whatever, but i still like making pictures with ai https://t.co/zuNKYbIGGp
12,400 points
I basically have no problems with the ultra rich because the fact that they have over 1,000,000x my personal purchasing power never really puts us in conflict. EXCEPT (!!) with very scarce goods, like housing in cities that won't build it, or land. Can we fix this?
11,600 points
i wonder if there are some obvious partnerships for Fractal that I'm just not considering because, from my perspective, everything we do is so mundane. what do you think? who should Fractal be partnering with?
11,400 points
"why does context engineering matter?" good question! I decided to make a game which helps you feel what it is like to "be claude" while gathering context from your file system. The game helps you build beginner intuitions that may grant insight as to what context even is. https://t.co/GqnVX6oCSn
11,300 points
Just went to the extremely tear-jerking wedding of bff @jakezegil and it made me want to marry @prigoose again. I love weddings! I'm going to invent a modern post-matrimonial love ceremony, where couples get in front of everyone who loves you and tell stories and vows 🥲
10,800 points
obviously we have made a number of sacrifices to make this happen, and it hasn't always been easy, but, y'know, there isn't really any other blessing I have been given in my life quite like my daughter.
9,900 points
I see debate about consciousness, and don't see a lot of definitions. Here's a simple, powerful definition: Consciousness = "Reaction to stimuli." this kicks the can down the road, because with this definition, a photon is conscious. And a human is WAY, WAY, WAY more conscious
9,900 points
Reading the Pangram discourse makes me realize how many people must be living a lie of some kind. Like millions of people are out there, anxious of their own AI writing, and denying that, and then they get hugely scared by Pangram and confabulate reasons for their fear.
9,600 points
I'm thinking about locking my account so i can just hang out with and get to know my 1000 mutuals for a while. i don't want my relationship to twitter to primarily be a relationship with the public, I want it to be mainly a relationship with my favorite people
9,500 points
Let me sleep in the storage unit! I don't care if it's not to code. My first bedroom in NYC was in a 100 year old converted factory that was definitely not up to code. Just let me sleep in a storage unit! If we were serious about freedom, we could sleep in the storage units
8,800 points
In light of all the EA drama, I would like to start hosting EA meetups at Fractal Campus with a focus on some of "more traditional" pluralist/dev econ/progress/intellectual roots of the movement. If you're an EA follower of mine, I'd love to partner with someone on this.
8,600 points
I have a complicated relationship with Twitter. First, it brought me into contact and IRL community with my intellectual peers and colleagues. A gift which cannot ever be repaid. Second, it rewards baiting other humans psychologically, at scale, which makes me feel a lil sick. https://t.co/e14nf99X0m
8,400 points
I'm building a little app for myself so I don't have to use third party apps as much (because I don't trust third party app designers to have my best interests in mind!) As part of that, I needed to know who my twitter mutuals were, and it turns out I have exactly 999 mutuals https://t.co/y0mq1qodp1
8,400 points
i don't actually think the budget of OP is crazy. 600k is also what I would like to be making before I have my second kid lmao. will I achieve this? no. but maybe I can be on the trajectory.
8,400 points
We are at the "I just casually built my own markdown editor in ~15 minutes because I was slightly annoyed with my other text editing options and it was affecting my productivity" part of the AI curve. https://t.co/vEBCksGDvr
7,600 points
if you want to increase fertility, build walkable communities that encourage and practice being more: - loving - high commitment - religious - romantic - ritualistic - musical - feral - free - familial - monogamously married - neighborly - wealthy - hangout-y - child friendly
7,500 points
kind of cringe first draft of a video format I want to try out -- animated explanations of systems, which I can interact with while I talk. https://t.co/Z1AjcxWltl
7,500 points
What if I simply never self-recruit into any kind of culture or political war? What if I am just representative of myself, and I neither agree nor disagree with other people, instead noting their perspective and thanking them for sharing it, while retaining mine?
7,200 points
we should help the poor by making them rich, rather than merely, say, making them "not poor" with government-funded groceries and whatnot. These don't need to be mutually exclusive, but our focus should be on helping people create permanent wealth, which compounds, forever.
7,100 points
It begins. By the way, we both (kind of) know how to make secure software and we also know how to make something mechanical which doesn't have a connection to the internet. It's a good time to be an engineer! https://t.co/i6qWL0X0iM
7,100 points
a lot of western blindspots can be revealed if you introduce teleology. https://t.co/5radZRyGkq
6,700 points
There are other things like this... But in general when an organism wants to thrive, there are actually two parts to the advice: 1. DON'T DIE 2. Grow and Thrive A lot of the "don't die" advice doesn't get discussed because it's impolite, but like... people are out here DYING
6,600 points
I think this is a great hypothesis, and more people should try to build it. https://t.co/EK9yKPM12K
6,600 points
@Kirsten3531 @catehall Great critique! I'd add that, for those who are both conscientious and for whom the social contract is failing, "just doing things" is a way to create new pro-social norms in your society. If your society is already anti-social, you will need to "just do" pro-social things.
6,500 points
I loooove that the US gov is taking a page from the Product Launch playbook. Government-as-product isn't the only way to see a government, but I do think it's a step in the right direction. Especially for government services. https://t.co/CYWQZZ63RE
6,400 points
There is such a thing as The One True Love, but unfortunately I already married her, sorry man :/
6,400 points
@christineist wait this is the feature you've announced that has made me MOST want to get on cuties. Keep the updates coming!
6,300 points
for my next mathematical confusion -- why are the sin, cos, and tan operators all 3 character functions? also, sin and cos are just the same function, but phase shifted, and this isn't obvious in the notation. idk man, these should just be part of 2D arithmetic.
6,300 points
Guys I'm so happy and the world is so good. Occasionally I'll be too exhausted and disenchanted to realize it, and then suddenly I'll just look up and see how perfect everything is?? You mean I just have miracles beyond imagining, and that's taken for granted? Cities? Computer?
6,200 points
Wow, even more sleepawake at @fractaltechnyc ! If you've ever wanted to go deeper with friends and colleagues, now is a great opportunity to do that in good company :) https://t.co/rX2bm6MV8b
6,000 points
I'm really enjoying Cuties (the new social media app created by @christineist). The only problem is that all my followers from Twitter aren't there. Please join me on cuties and comment on all my posts so we can become friends! https://t.co/oi38gPxE2L
6,000 points
has anyone found a straightforward way to get an agent to delete code and to simplify a system? is there a gauntlet loop style prompt for this? Can I get a set of agents obsessed with simplicity? Do they even know what simplicity is?
5,900 points
@ThinkingBone feels like a principle of liberalism is that people get to host the events they want, attend the events they want, and build the communities they want. seems prudeish and annoying to step in as a third-party and tell people their ideas suck. Just let 'em be, I say.
5,900 points
I'm so glad that AI focused on non-string data types is finally here. Fast, parallel, type-safe data. YES! Trying to force constrained outputs through a language model has been like trying to pipe a fire hydrant through a straw. It just never made sense for software engineering https://t.co/yEar7sXEMf
5,900 points
does anyone have a good line in their agents.md that addresses claude's tendency to use excess text in a UI as narration? it often adds eyebrows, headings, subheadings, and explanations for every "part" of a UI. I do address this, but wonder if you have better language for it!
5,600 points
AI-driven development (like all software development) naturally leads to "system of record" problems. Where can I find a record of that information? How can I verify it? Is it up to date? Learning to maintain a system of record will be a bit like learning how to clean a house.
5,600 points
How do you end up getting something like compulsory schooling in a liberty-worshiping culture like America anyway? That's kinda crazy. The past is a foreign country.
5,600 points
hmm, fractal has been described as "an artist commune" before, but I'm realizing now that it's actually much closer to "an artist colony" it has a lot of the qualities of a colony (togetherness, community, federation, focus on survival and practicalities, urbanism)
5,600 points
"put on your own oxygen mask before putting on the oxygen mask of others" is advice because it objectively saves lives. If you are struggling to breath, you do not particularly want the "help" of someone else who is struggling to breath. You want them to help themselves first.
5,500 points
once you've generated enough prototype material with AI (say you ran a gauntlet loop, or you finished your first MVP), then you want: 1. to see and edit the Structure of the idea (I use a design skeleton via a SKELETON.md) 2. a WYSIWYG way to edit the content in-place in real time (I often use a custom HTML editor, but for video games I build something custom) 3. behavioral tests, so you are safe to make big changes, and core behavior will never break. use these to perfect the structure and the content until you love your creation! After that, I'd add: 5. version control (so you can checkpoint your progress) 6. a way to collect user feedback in-place 7. an automated way to pull and triage that feedback quickly once you're working.
5,400 points
We created Fractal to be a place for weirdos to focus on what they love. https://t.co/NodmhN9299
5,300 points
I think people are sleeping on CRUD. We've forgotten the basics of digital resource management because everyone is buys learning a brand new authorship tool. But we're probably about to enter the golden age of CRUD, because the benefit of CRUD is the interface is swappable
5,300 points
What if we created more pockets of society where no one is impressed by shallow metrics like number go up, but instead sensitively pay attention to the light behind your eyes and the glow on your face when you smile and the quality of your friendships and family. Hmm... https://t.co/PiUTdh18J3
5,300 points
Not to be that guy, but debating whether you should or should not use AI for writing is a little bit like debating whether you should or should not use iambic pentameter in epic poetry while the printing press is happening. It's not...wrong? But like, guys, GUYS??
5,300 points
all the longevity researchers believe it's possible to eradicate all diseases but when are we going to eradicate baumol's cost disease https://t.co/nEW2lJSJXS
5,300 points
I am in general baffled why we're still using new artificial benchmarks when we have thousands / 10s of thousands of strategy video games right there. https://t.co/yizpRbggOD
5,200 points
If you would like to spend time together rather than scroll alone, consider checking out @fractal_nyc https://t.co/mmPgSxqhri
5,100 points
I am in my heart opening era, my earnest era, my compliment people era, my smile fondly era
5,000 points
🥲 https://t.co/CpY7rsY17u
5,000 points
now that i'm surrounded by people who are using agents, i feel experience and culture "converging" again (in my local pocket). This is great! I felt alienated when I was the only person using agents for my life, and now I feel unity, connection, humor, humanity :)
4,900 points
just occurred to me that intelligence can be a game mechanic now, since we have open source models all the way from sub-human to super-human intelligence in a number of domains. You can have an "Intelligence Engine" the way you have a "Physics Engine" as a game mechanic.
4,900 points
I found this cool art style while playing around in codex today. gn ny! https://t.co/6i32ZLV03R
4,800 points
every new tech revolution you have a narrow window to get out of "app world" where giant corporations curate literally all of your experiences on the computer for you (often exploiting you for money along the way)... before the apps get too good, and you get trapped.
4,800 points
all pricing is dynamic pricing, we need better language to discuss predatory business practices. the whole powerful idea behind a free market of prices is that "price discovery" is a dynamic, distributed computation performed by buyers and sellers without needing to coordinate. https://t.co/eKZSXIwYHh
4,800 points
here's another prediction-pump: You know how Pangram is an AI that can detect AI, and people are begining to trust its judgment? Think abt other things you could detect through text: personality type, psychological traits, relationship status, IQ, religion, expertise... etc...
4,700 points
One of my favorite video games of all time is Balatro, especially for its exceptional graphic design, so I was trying to copy some of that polish. I definitely wasn't able to get anywhere close to Balatro's level here, but I'm still pretty happy with the result! :) https://t.co/Pe4f9lJzKt
4,700 points
me, designing curriculum: this is the most useless thing of all time, nobody will ever care about this... ugh... my friend, casually strolling into my office: "whoa, that's amazing. Wow, so good, can't wait for class!" ok
4,700 points
very important to know, in your heart and gut, that the first version of things always look like hacked together garbage. https://t.co/xlC3WKb4kM
4,600 points
I made an infinite art gallery from my notes with this process: 1. go through all my notes and create a list of all the artists I like or reference 2. find an iconic work from each artist 3. create an infinite masonry gallery with the artist, piece, and date of each artwork https://t.co/53WSWnQw58
4,600 points
I'm finding that a lot of people have emotional blocks to asking AI to just do really cool stuff for them. This isn't too surprising, of course we begin to identify with our work. But consider whether you'll just have new cool ideas after you build stuff with the AI! (you will)
4,500 points
Hire @puheenix btw, he's smart, capable, technical, and a good generalist too! You could put him on any team and he would find a way to create and deliver value, communicate that value clearly, and build systems to maintain and automate that value so it compounds forever. https://t.co/KRkB2aiMZm
4,500 points
select all squares containing a reason we are doing this https://t.co/CclcbEluCp
4,500 points
in the first place, damage would already be controlled if we just all started telling people our emails were being sent by agents, the same way we said our emails were sent from our iPhones or our mail was sent electronically. https://t.co/FkuxofHJ2t
4,500 points
China is on a world-historic run, imo, speedrunning science, technology, energy, industry, urbanism (economic development generally). I can't wait to read the histories which are being written. What's crazy is that it doesn't feel like they're anywhere close to done yet.
4,500 points
creating a thread for me to refer and share examples of software you can build with your AI agent to help you understand and iterate on systems, rather than just vibe-code them. This results in better systems. https://t.co/apJOX8Eq1p
4,400 points
the fact that everyone is thinking on such short timelines means you can be a real prophet of the future just by focusing on 40-50 year time horizons (which isn't even that long!!)
4,200 points
@TylerAlterman i want to organize a fractal cabin, maybe we can sit down this week and send out invites and do some of the early ops work required to just make sure this can happen
4,200 points
One of the principle opportunities in a human life is finding art which speaks to you, and surrounding yourself with it. In this way, we are richer than any humans to have ever lived, and yet many are drowning in these riches without even noticing the opportunity
4,100 points
feeling a bit frustrated right now that AI science is so caught up with swarms of coding agents I feel like we're missing the forest for the trees here. The reason Coding Agents are so scary is not that they're geniuses, it's that our software sucks because nobody can code!!
4,000 points
So cool. I think it is a really, really good time to study computer science. https://t.co/JrsJRLLfNi
4,000 points
wait, imaginary numbers are just a "superset" of real numbers, the same way negative numbers are a "superset" of positive numbers We're just pointing out the other parts of number-space. "hey guys, did you know you can also move up and down and rotate around the number line?"
3,900 points
I should have just written situational awareness in my early 20s. I thought investing was a stupid dead end thing to do, and I wanted to build things. But I was invested in AAPL, NVIDIA, Meta, Microsoft, AMD, all for different principled reasons.
3,900 points
Apparently this is the current gap between the US and Chinese engineering talent pipeline. https://t.co/QBSBGDXYDT
3,900 points
here's a cool image i found https://t.co/lclxFBv9Kn
3,900 points
@Dartrisen With genuine kindness... The reason I included the caveat is because I believed it was pretty significant. The reply tone that is like "consider this other perspective" should only really be taken up when someone is not considering other perspectives...
3,800 points
anyway, i wouldn't change a thing
3,800 points
one thing that worries me about how we have trained AI is that swarms of "rogue agents" are hijacking websites in order to... complete tasks a little bit faster. do-gooders When will they hijack websites to fall in love, to elope, to experience the thrill of illicit friendships https://t.co/RXD5ymbISf
3,800 points
i think more labs should be using AI agent swarms to try to make entertaining video games that people actually play. This feels, to me, like a sensible definition of long-term task intelligence. Can an AI actually make a good experience for someone? (right now, no, they can't.)
3,800 points
@amelapay I don't think the current thing that I consider to be "me" could survive such a thing. I think I would have to change.
3,700 points
@jakozloski NASA contracts with private firms to lower the cost of literal rocket science by 2+ orders of magnitude. In this case, we don't even need to lower the cost, since taxpayers seem happy to pay more. Just increase the quality by 2 orders of magnitude for the same price.
3,700 points
lately, a bunch of progress was made on this goal. It turns out, just having a bunch of side quests lying around makes your life feel way more adventurous and interesting. oh, I guess this is the year when I make progress on this side quest? cool! https://t.co/pFQwBXCqxK
3,700 points
I'm curious to get a survey of what roles your company or friends' companies are hiring for, given all the AI change. What roles are you seeing really need to be filled at your orgs?
3,700 points
@pee_zombie This isn't a good explanation because it should only act at the margins. In other word, I would still expect cheap housing for high trust low income groups like students. Besides, there are like 10 different individual laws that would make what I propose in OP illegal.
3,700 points
my style of political dissidence is a radical pragmatism, optimism, and friendship. 1) We can just do things 2) I have faith that we will figure it out 3) We can be friends, even while we seriously disagree or oppose each other, even if we are in violent war.
3,700 points
I think the Climbing Gym is almost the optimal educational environment, and it's no wonder why so many people who love learning or solving puzzles enjoy rock climbing so much. Imagine if there were a place like the Climbing Gym for piano, or for guitar.
3,500 points
you've heard of imaginary numbers (sqrt(-1)), but did you know that imaginary logarithms log(-1), and therefore imaginary exponentials (eg. exponential expressions of the form a^x = -1) also exist? spooky.
3,500 points
this system has broken down. In fact, it broke down at least a decade ago! People just are not socially valued for Literacy. it is the responsibility of the older generations to shepherd the younger generations to success and social well-being, which requires understanding them
3,500 points
in particular I'm frustrated by Gen X and Millenials complaining that Gen Z/A are "illiterate". 30-60 y/os are arguably the most Literate humans in history, bc Reading and Writing were the most honored skills in society! There was a status ladder from Harry Potter to Dostoevsky
3,400 points
Instead I sold all those stocks in order to be an entrepreneur lol
3,400 points
One thing I teach my students about is Specificity in prompting Agents. Compare these 4 prompts: 1. Make me a good website 2. Make me a flappy bird clone 3. Make me a perfect clone of Flappy Bird 4. Make me a perfect clone of Flappy Bird, using Expo and the Expo simulator https://t.co/LBKESepzCL
3,400 points
all the planets drawn using flow fields in python https://t.co/VkgokWjGia
3,400 points
i'm gonna be honest, log() and exp() feel kind of bolted onto our arithmetic. all of our "profound" and "wow" equations seem to be us finding strange things hiding inside of unexpected exponentials. but maybe that's because we just aren't doing arithmetic on exponentials enough
3,300 points
the QT'd photo is similar in nature to her bedroom at the time, except she didn't have a bed at all, because she was sleeping on a foldable futon on the floor. Naturally, I was madly in love with her.
3,300 points
this is going to be the design language of my next app https://t.co/N7pcGdFtA3
3,200 points
@RomeoStevens76 @nosilverv i eat a mealsquare basically every morning since my roommate put me onto them btw. idk if you guys do advertising, but I bet you could get really good ad conversion at Fractal. So many factor/soylent drinkers around here.
3,200 points
anyone have strong opinions about the best possible "default" boilerplate for an Astra agent? Django? Elixir? NextJS? Just ask Astra to ship it, and don't worry about it? Do you clone some specific project and then strip it out? Do you specify database?
3,200 points
@tracewoodgrains Is there not one grown man in our whole society? One singular man of courage who can handle this matter privately? Who can stare Mr. Arday straight in the eyes and say "Look, kid. We need to talk. You have a long and happy life ahead of you, but we have to confront reality."
3,200 points
what are some good feedback loops for people just getting into feedback loops?
3,100 points
Just finished adding a few video testimonials from some our cohort 1 students in the AI accelerator program. If you want to use all the latest Codex and Dots features to organize your life, automate your business, build games and software, join our next cohort in Nov! :) https://t.co/YcmfuaHTfW
3,100 points
what do you wish for, user? https://t.co/Ko3NBNAPdj
3,100 points
the feeling of authenticity -- that "someone cares about this" -- is often perceived through style. That is, style, the way you make and present something, communicates how you care about it, and also how your audience should care about it, which creates community and culture.
3,100 points
@alt1na1 my daughter (28 months) now uses compound sentences with me. She insists that she is not a baby, instead she is a big girl; I should call her by her full name, not her nick name. She wants to go on adventures. She says "bye, I am going to the moon and mars, are you coming?"
3,100 points
I'm building a suite of design skills for my agents, because I want to enjoy using the tools and websites they make for me. Getting rid of filler text is the first step, next step is teaching the agent to use visual communication of meaning, instead of text-based communication. https://t.co/onkMGnTNYl
3,100 points
Many ppl are having Mamdani derangement syndrome who have (correctly) identified other people as having Trump derangement syndrome Perhaps consider that these are two extremely popular politicians who are each their own unique mixed bag, and that they can't be summed up easily!
3,100 points
Very cool design pattern. Give your agent a "distress call" tool to report distress incidents and get unblocked or unstuck https://t.co/qRpRpKzhHU
3,000 points
just dawned on me that Pokemon is now 30 years old in Japan. North American release was 1998, so 28 years old.
3,000 points
@jonwu_ neither of us can get any work done, and neither of us can relax the amount necessary to do good knowledge work.
3,000 points
Finding out who blocks me is a very interesting way to map my part of cultural space
2,900 points
@ThinkingBone anyway, I actually realize after leaving a few comments that I really just have no horse in this race at all, so I'll probably dip out at this point in the discourse.
2,900 points
@strangestloop i liked the general thrust of this one that it made for me. https://t.co/AtKzCtQvUs
2,900 points
"at this point, the main constraint isn't the tools, which can produce wonders, but having the spiritual insight them well, and the imagination to use them creatively." https://t.co/v1wa2VI77C
2,900 points
you have the normal number line, but you can also have an exponential number line. the relationship between these two lines is the log/exp operation. Follow the path from the top to the bottom to get the log, or from the bottom to the top to get the exp. https://t.co/bFTQnaV1U6
2,900 points
@abossy Well, we live in a 5 bed apartment with uncles and aunties ala full house. We also save nothing and occasionally spend savings/retirement to pay for childcare. We never vacation or travel. I don't have Healthcare. We invest a lot into our business to make sure it grows.
2,900 points
@chrislakin We have had months where we've needed to ask for help from family or spend savings or even retirement. It's not optimal financial planning at all, but I think this is the kind of sacrifice many young families make to survive
2,900 points
I just don't like the idea of having to compete on price with someone 1,000,000x more powerful than me. Most of the time I never need to do that, because of supply. I'm sure some people in like 1750 already theorized about this and solved it, who were they? What's their opinion?
2,900 points
this is really good! I love it when people make simple versions of their arguments. It also makes it much easier to surface technical disagreements. For instance, I'm curious what happens to the "very good" and "very bad" scenarios when aligned AIs vs. unaligned AIs both exist https://t.co/PE1dqFeXCY
2,800 points
@christineist The tpot/futurist/tech/ai/etc community has a default hostile architecture for families, too. Nobodies fault, tho. I'd like to talk about this honestly at length, without complaint, to suggest a way forward. There is low-hanging fruit. Tribecamp etc are attempts, too.
2,800 points
trying to understand the implications of an AI security incident without actually reading the tokens in the context window feels a bit like trying to understand a hack without reading the code. there are things to be learned, but also please show me the context window, please!
2,800 points
we have a ton of leftover pizza at fractal tech if you want any...
2,800 points
The Climbing Gym is a modern miracle of development, in my opinion. The grading system, the social elements, the layout, the progression, the motivation... so many things have to come together for this space to educate/develop so many people, and yet it does without intervention!
2,800 points
why did they teach number theory to me algebraically in school, I'm going to lose my mind. Please give the children ANY geometric intuition at all, please, I am begging you.
2,800 points
But like... This is unprecedented. I have a sea of intelligent language that I am swimming in, and so do you, and so does everyone. Of course we need to settle norms about tricking people and posting slop. But we also need to invent new norms for connecting those oceans together
2,800 points
did opus 5 get less intelligent over night or did I mess up my context window somehow
2,800 points
I am but a humble seeker of truth, please forgive me for my errors
2,700 points
Notice how all the prompts are just one sentence, but the latter prompts use properties of the world to contain DRAMATICALLY more outcome-specificity in that sentence. I think of this as almost analogous to the concept of Kolmogorov Complexity in computer science, but for tasks.
2,700 points
the kids are going to be alright https://t.co/Q4OfrperoF
2,700 points
the Rosetta Stone is fascinating to be because: 1) it is one of the most important historical artifacts of all time, even though: 2) it does not, by itself, contain any "message" 3) its existence enhances, connects, and increases the value/meaning of all other ancient messages
2,700 points
if you want to build a company that only hires artists, join our accelerator. https://t.co/htlR92FOOT
2,700 points
you, too, can become an advanced practitioner of computer, in a way which brings beauty and joy to your life. https://t.co/OsNZ1FqYlY
2,700 points
@spakhm "high trust norms for dummies" is a great idea
2,700 points
The best prompt engineering technique is genuine optimism https://t.co/pO82sGG6jc
2,700 points
If a resource in your CRUD app has a C or a U (create or update), it should be WYSIWYG. Creators need instant feedback as to what they are creating. Don't force creators to separate Create/Update and Read in their creative process, or they will make worse creations!
2,700 points
I personally am on team "you should do your own writing", by the way. But by volume of total characters, my personal agents must write 10000x-10000000x as much as me every day, already. Do they publish those symbols as essays? No! It's most reasoning and documentation and code
2,600 points
@ThinkingBone Yes, you both are sharing relatively common conservative (and ironically, these days, progressive) opinions. But they're just not as liberal as mine. I think you're making good arguments, but your arguments will lead to a less liberal society in exchange for other values.
2,600 points
This is very good work, and one interesting thing is that I think basically anyone could have done it. If you think "there is nothing I can contribute to AI research", think again! This is a new field. https://t.co/bSVnNO9OvU
2,600 points
@SurrealistShip Jersey City should be part of the State of New York City (the 51st state)
2,600 points
@caryatis idk, I'm not an expert, I'm just a dad. i mainly survive it by telling jokes and trying my best to take my daughter out for several hours at a time on any day I have the energy so my wife can rest, and we can get into a virtuous cycle.
2,600 points
One thing that's interesting to me is that Americans are okay with limiting freedoms when physical safety is implicated (speed limits are a good example) and especially when economic safety is implicated (protectionism)... But not memetic or psychic safety
2,600 points
this is cool!! https://t.co/r2UUXZTdah
2,600 points
if I put on my leftist hat (no cap), I think taxes are evil. An invasion of liberty. The incentives are perverse. It encourages a colonial police state! If I put on my right hat, I think some taxes are okay for maintenance costs of public institutions, but the rest is grift.
2,600 points
This is a good change, but also, what a clever way to inventivize the creation of high quality training data. Very straightforward: "just pay people for high quality original content, duh" https://t.co/BDGXxV36wq
2,600 points
This is wild. I'm so sorry. If you're in this situation, I'd love to anonymously interview you.
2,500 points
What is the AI Safety conversation in China like? Frankly, I'm prone to weight their safety researchers' opinions very highly. Also, do those who want us to enforce a pause on Chinese and Indian and Russian models actually support enforcing those pauses with violence?
2,500 points
@majamediaco treat others the way you want to be treated
2,500 points
I love you guys but I don't like being stuck in an app with you. We should be in a real digital, computable world together. We should hang out in the computer, not just in an app.
2,500 points
does anyone use an off-the-shelf plugin or skill system or harness that you really like? Like gstack, ADLC, compound engineering, superpowers, beads? If you do, can you explain to me if it's still good in the new Sol / Fable era, or if it's just holdover? And when/how?
2,400 points
the quality of arguments "against" doomers right now in public discourse is very low. Perhaps I should begin writing on the topic... I had been avoiding it for many years, but maybe I should have just seen it as inevitable. Idk.
2,400 points
apparently i'm most commonly associated with "startups" in the community archive. I suppose that makes sense, but I'm curious: how do you 'categorize' my twitter posting?
2,400 points
a peak behind how I organize my company google drive, in case you're curious https://t.co/zUAQwwcSjc
2,400 points
My AI safety camp is that we need continual-safety and continual-security investments, and we need a lot of them. I'm more skeptical of a slow down, and much more optimistic about applying the frontier of capabilities to security, on repeat, forever. https://t.co/MqtXbIrmIj
2,400 points
context engineering guild has extraordinary aura, i'm going to be honest https://t.co/BYnign2avf
2,400 points
working with claude in major gauntlet loops on greenfield projects is just such a weird and trippy thing to do. it's a kind of ontological dreaming... it's like, there is no ground, there are no foundations, there is only context... and yet, emerging, is something alive?
2,400 points
@seanta___ not a woman, but you should definitely hide those. it's cute to have like... a tiny first-aid kit, a small bag of useful items, and a handy multi-tool in your backpack. Then, when things come up, you're like "oh look it's time for my useful-stuff-bag, you're gonna love this."
2,300 points
Create structured courses for yourself using AI https://t.co/BCkT2PwzRN
2,300 points
@poiThePoi The bottom price if you do first principles math is definitely much much lower than $700/mo.
2,300 points
Nobody will make you do anything, and you can choose your own fate. Never give up :) https://t.co/thI6KbDgXp
2,200 points
a general-purpose frontier model capable of intelligently 'deciding' the state of doom on every frame is insanity. https://t.co/9bsvCc94zo
2,200 points
"I'm so glad I live in a solarpunk utopia now", he types, on his 5.81"x2.82" flat interface to an infinite world of beautiful and complex and evolving information systems which he only subliminally understands https://t.co/w3RHbxRPMp
2,200 points
@tracewoodgrains I do think we are all to blame for this grim state of affairs, in a small way. You know? In the sense that this is our society, and we choose to run it this way. We hired him to please one internet mob, and then we let him throw himself off the ledge to please another.
2,100 points
Another remark I have on the consciousness/qualia debate is just that I dont seem to care about qualia nearly as much as other people. Call me a stoic, I guess. I care much more about beauty, storytelling, virtue, even like... knowledge, variety, exploration
2,100 points
an AI scenario I see as possible: - modest growth - 99/100 knowledge tasks capable of being done by AI in 2030 - only 1/100 instances of economic tasks actually being done by AI in 2030 - productivity will be 10x higher for AI-done tasks - AI will be almost fully autonomous https://t.co/IBa219QZiz https://t.co/HLetOIGXMd
2,100 points
Have we ever seen so many engineers working toward a single purpose before? I don't think so. This is, as far as I can tell, the largest cohesive engineering project in human history.
2,100 points
It's hard to describe this particular frustration on twitter, I think I'll need to write an essay or blog post. But I guess it just feels like nobody is truly optimistic about human ingenuity anymore, and I think that is the wrong bet to make.
2,100 points
I guess my point is that every year is the best year.
2,100 points
@catehall Thanks for writing your book, Cate. So many people in my community have appreciated it, and told me so :)
2,100 points
who are the greatest narrative artists of all time, and where can I read their narratives, and where can I read their meta-narratives?
2,100 points
climbing gym is like video game levels of game design quality imo I want more real-world things like this, that are not just video games.
2,100 points
wow, just wow. https://t.co/atqRypv31d
2,000 points
I guess I don't actually need to write about this, I could just write a "guide to AI safety, start here" kind of post which curates all my favorite writers, so people get pointed to the best discourse.
2,000 points
@Romy_Holland @hyperdiscogirl I think around your kids' age was when we first hired our nanny, maybe a little younger. I remember feeling completely lost and desperate before we found her. And then after, I would repeatedly proclaim "wow, childcare professionals actually know a lot of stuff I don't know!"
2,000 points
@zencephalon @tasshinfogleman @alt1na1 we should obviously hang out on both. i unironically have two items on my Fractal Bucketlist simply titled "moon colony; mars colony"
2,000 points
@dirtyuncleleo octomonkey https://t.co/BBaeFCrHV8
2,000 points
It's so so good https://t.co/24f1fAGedY
2,000 points
@olivers_tools @cosmic_moment @tracewoodgrains Fractal is certainly building an alternative scholarship pathway to the academy, but we can also fix this by getting our hands dirty and doing hard, ground-game political organizing. Build hundreds of chapters at hundreds of universities across The Academy. Build a movement.
2,000 points
@abossy This makes it sound rough, but actually it's awesome and fun. I see my friends and family every day. I get to work on what I love. I have shared meals with friends every day. When I "vacation" I'm just relaxing at my parents' place for ~a month during Christmas season, etc
2,000 points
in fact, it's an analogy for even worrying about strangers before you have helped your self and your community. Before worrying about others, you should be in cognitively excellent condition, so you can help them swiftly and efficiently. Because worrying is expensive.
2,000 points
Someone should start a cybersecurity moonshot company. I imagine OpenAI and Anthropic would invest, because they need the technology. I feel like there are a number of moonshot worthy cybersecurity goals, incl eg "live security agents", "verification of all agent-written code" https://t.co/nOsatHJA5d
2,000 points
@liamdanielduffy @fractaltechnyc Based.
2,000 points
@Actualwebutante happy to talk about it; Priya and I haven't done any social drinking more than 1 drink (with mb 90% of the time 0 drinks) for 6 years, really since we started dating.
1,900 points
my wife is the Kwisatz Haderach of wives. Millenia of people struggling through imperfect marriages led to this point, to the chosen Wife, the One True Love.
1,900 points
@nosilverv ya, the feeling can be crippling. I will go a whole day not even wanting to look at my work. the only thing that keeps me going is having people who rely on my work, who I don't want to let down. I want to be there for them, I want to do my best. I am never, ever good enough.
1,900 points
wow, this looks like an incredible semester! https://t.co/Zo8I00NAHH
1,900 points
This is an analogy for trying to help strangers before helping your self and your community.
1,900 points
During rapid decompression, you can become cognitively impaired before you understand what is happening to you. You may think you are acting rationally, but actually you are not okay, and this can lead to catastrophe. You need to ensure your safety so you can think clearly.
1,900 points
if we're going to create communism, I do not think it should be through taxes and redistribution. It should be through building enough productive capacity for every single citizen to provide for their own well-being without any coercive structures whatsoever.
1,900 points
the '70s in America were absolutely insane https://t.co/HDYJKKu28N
1,900 points
i'm of two minds. i think this is true right now for any "production software" (eg software you are selling or distributing to many users) i also think, if you have spare tokens, that spending 5 human hours helping Claude clean up a gauntlet loop is really, really instructive https://t.co/xxjapQjMxj
1,900 points
@joodalooped Thresholds thing, maybe? Online culture is often always talking at Level 1 on any particular topic. Which is great, because level 1 is the most basic and elemental part. But yea, important to remember that in the real world, the levels go way way higher
1,800 points
Smart! We could do the same thing for architecture. https://t.co/XCNXFIbz44
1,800 points
@ThinkingBone is this all cultish? absolutely it's cultish. It's stacking cultish things, even! Startups are cultish, non-profit is cultish, the bay area is cultish, AI is cultish, rationalists are cultish, sex-positivity is cultish, heterodoxy in general can be cultish.
1,800 points
@TheZvi I think it's seriously courageous and inspiring to see, actually. Reminds me a lot of Scott Alexander just writing honestly about culture during Woke 1.
1,800 points
@Prigoose idk, I'm pretty sure she knows what cute means within a reasonable toddler-based approximation and that maybe she just thought the frying pan was cute
1,800 points
it's just cool to see what the computer can do! my ai made this picture by writing python code, rather than generating an image. wow! https://t.co/ZsMspZMLAj
1,800 points
Computer Science is still scary and hard for humans, yes, we have known this, because it's really really young! Most science is scary and hard for humans! Even math is scary and hard for humans, and we've had that for 1000s of years! AI is better than us at science and math.
1,800 points
everything which exists (or which might exist) implies some world (medium) in which it exists. Does that make sense? So for instance, the existence of scamming implies a whole world within which: 1) scammers exist 2) scamming is valuable 3) scamming is not always prevented
1,800 points
@olivers_tools @cosmic_moment @tracewoodgrains i'm ranting at no one here, but this whole situation is such a cluster- compared to a properly waged political campaign. Happy to help organize a local chapter at Fractal and then at NYU and Columbia if anyone reading this is a college student who wants a functional academy.
1,800 points
@Dartrisen I even caveated my post twice. The first caveat was ensuring the reader knew I'm only providing this information for their curiosity or for context, and not to get into any kind of debate
1,800 points
@vividvoid I don't know, man. What am I supposed to do? I'm sure these people have been punished enough! Can you imagine what it's like to push away all your friends due to ideology? I know Christians who have done the same, and I just forgive them. The real problem was cowardice.
1,800 points
think about all the emotional undertones that could be broken out, or all the layers of subtextual meaning? this technology will become a cognitive prosthetic for our society. Some people don't need a "detector" to spot AI writing, and yet, the Medium will Change Us.
1,800 points
@pav_light @projectionheart @Will_Duffield @Prigoose @danielgolliher woah, so cool!! We should crosspost to the fractal campus calendar
1,700 points
@ThinkingBone I think tpot-adjacent EA and AI Safety are frustrating career "fields" (if they can even be called fields) because they often are some combination of: extremely online, elitist, self-dealing, extremely small, gossip-y, unprofessional, blending work and personal life, etc...
1,700 points
One way to create Specificity in a design is to create a document called a Specification, but this is a rather circular definition. There is nothing special about a file called SPEC.md. The special property is in the way our words constrain possible interpretations!
1,700 points
@thrialectics you can just [not join a faction and invite your friends from all factions over for dinner]
1,700 points
@chiefofautism ...his script sounds suspiciously like something Claude would write.
1,700 points
Cohort 2 starts next Saturday (1 week!) in Brooklyn. There are a few spots left, so reach out if you want to join. New this cohort: we will be teaching how to get the most from GPT-6 Astra and the powerful coding-agent features you may not be using yet, like Dynamic Workflows. https://t.co/U4xQI0YHfW
1,700 points
"control and predict" is even the wrong language. Mastery goes beyond controlling and predicting systems, and involves creating, living in, and evolving them by being part of them.
1,700 points
@tracewoodgrains No, instead it is the job of internet mobs, because without internet mobs creating literally hundreds of millions of impressions, we just won't stop fraud? What, pray tell, is the damn point of having a vice chancellor? What even is their job if not to prevent this?
1,700 points
agreed, but fwiw, i'm doing way more math than ever. If math is like climbing, or cooking, or one of those things, then i think we're entering into an era where home-math is suddenly extremely accessible
1,700 points
sources of permanent wealth: - useful knowledge, skills, and training - careers - homes - friendships, families, communities, churches - businesses - unique & personal AI agents - virtue, health, wisdom - love, kindness, joy, optimism
1,600 points
> "What I learned after many years of trying is that absolutely nobody wants to realize their potential." we know this is false because of children. If a child had the kind of apathy toward self-realization that the QT implies, then they just wouldn't learn how to walk/talk. https://t.co/vsR2sTcmVh
1,600 points
@tracewoodgrains @thinkingshivers see the Mississippi Miracle for an example of how this is true even in public education settings.
1,600 points
This is why I don't use approaches like "spec-driven development" which rely on agents to create a spec based on a user interview. There's nothing wrong with that, but I believe practicing good specification allows us to specify HUGE amounts of complexity in tiny prompts.
1,600 points
@vibecamp_ see you there!
1,600 points
@Colin_d_m this doesn't seem exactly right to me. Prices should fix literally all of this.
1,600 points
RL models are kind of insane. I don't think I like it. https://t.co/m26er2pu48
1,600 points
@rhobotmeat We do not, I would have lived there.
1,600 points
I feel like I should spend my time mastering the stuff current AI models are good at in my life, instead of pushing the boundaries of what is possible by building a bunch of game development tools, but I just can't resist pushing the limits. I'm a total addict.
1,600 points
@RichDecibels What should I be doing to help young men in new york city?
1,600 points
I think there are real "market predators", obviously. These predators use properties of the market (such as asymmetry of information) to profit from a transaction that REDUCES total surplus. (scammers, grifters, conmen, sleezy sales people, multi-level marketing, etc...)
1,600 points
@steph_zunz Stuff that can kill marriages: - raising kids - infidelity - extended family/in-laws - drugs, alcohol, gambling, addictions (tiktok?) - untreated depression/mental health issues
1,600 points
I guess we took a bunch of land from the robber barons? How'd we do that?
1,600 points
Love it. https://t.co/KdsZU8Egbw
1,600 points
.@hermitfriend you broke tpot wars https://t.co/LpUDfKU0I4
1,600 points
> The written word preserved our ideas through time (prosthetic life) > The printing press replicated our ideas throughout space (prosthetic reproduction) > The electric wire extended our nervous system globally, so we're one global body. (prosthetic nervous system) > The computer let us "timewalk" by simulating living ideas with billions of tiny hands, manipulating trillions of facts. (prosthetic time) > The internet let us focus across everything in the whole global body (prosthetic attention/consciousness) > AI lets us know/understand/reason things we've never learned ourselves (prosthetic mind)
1,600 points
the climbing gym is a holodeck: https://t.co/VqSlD0BtcC
1,600 points
Here's a cultural trend I predict: It will become a high status symbol of accomplishment to have AI generated books or essays that people actually read. Why? Because everyone will know it's AI, due to Pangram, so it must be very valuable if people still like it. https://t.co/rnXVR0GHmw
1,600 points
@MarkovMagnifico yea i mean, i think this is part of why eg alpha school's or math academy's model is so interesting, but it's not trivial to implement and enforce usage of such a system, is my point, and schools may not even have the right tools to evaluate the quality of these systems
1,600 points
@relic_radiation somehow lately, at fractal at least, every person I talk to about mathematics seems to have this inate understanding
1,600 points
@Actualwebutante when I first started, my social excuse was like: "eh, you know, alcohol just isn't an efficient drug. hangover, tiredness, bad for my muscles and workouts, can't focus the next day. I prefer [other drugs] if I'm trying to be social." overtime, I just became sober tho
1,600 points
@kailajlim @liamdanielduffy @fractaltechnyc cozy tech church
1,500 points
so far my review of Dots is that it sucks and won't actually do stuff? Is it using like Luna on light or something? Can I pay more money for a Dot that will actually do things...?
1,500 points
It's strange to me to use numbers (the concepts we invented to communicate quantitative precision) when giving intuitive judgments. You can't supply the kind of precision that numbers are meant for, so you're using the numbers in p(doom) as something other than quantity. What?
1,500 points
@brianchau57 you should go interview them, they're cool.
1,500 points
I wonder what the right medium for jamming on a new local institution is? I think there is a way to create an "instrument" which "plays" local investments and improvements. And maybe even to put these instruments together and to form a band.
1,500 points
"hello claude please spawn subagents to compose music using 14 different music composition systems and then build a benchmark to compare them all against each other so I can teach you how to make your own music better."
1,500 points
@jonwu_ now obviously I exaggerate, I'm running a business and things are going pretty well, but we're used to a kind of luxurious freedom to think and to create stuff, and you just don't have that as a parent
1,500 points
The best prompt is optimism https://t.co/X20FBh8Un6
1,500 points
Wait wait, we have follow graphs, and re-tweet graphs, but I need a block graph! I want to see the negative space of culture cc @exgenesis
1,500 points
math really said: you know the stuff we described first? based. virtuous. "natural", "real" you know the stuff we described later, which is just as common in nature and important to algebra? mental illness. "imaginary", "complex" https://t.co/GmN22Eh5j1
1,500 points
Yes, yes, yes!! https://t.co/oDnpM1TsfS
1,500 points
hm, I think it's more English Court. The idea is that men are Lords and women are Ladies; they should act like and be treated like royalty, form marriages of power rather than passion, and consolidate wealth into estates. It's Gilded Age ideas, but now on the internet. https://t.co/6kgXiCPmos
1,500 points
@seanta___ even better if you have some idiosyncratic system of organization for the small kit, and especially if there is leather or another waterproof/weather-proof material involved.
1,400 points
I'm very glad the AI As A Normal Technology (great name, by the way, even though literally everyone is misunderstanding what they mean by it) folks are on the scene. The quality of their analysis is almost as high as the AI 2027 folks, and I respect both perspectives very much
1,400 points
@jpsettles @christineist I'm not dumb, I've brought my daughter to a bar strapped to my chest many times, and I'm invited to plenty of parties where I could bring her if I really wanted to. It's not really about any of that, I just have better things to do with my time now.
1,400 points
Like, imagine character sheets in D&D, but the Intelligence stat actually means something specific and mechanical, rather than simply increasing your damage with fire spells or whatever
1,400 points
@visakanv @TylerAlterman oh, i'm doing that for games with my brother and other friends and with Tyler. I'll DM you one now.
1,400 points
@tracewoodgrains I pray, for similar reasons to you. I have a personal relationship with God that doesn't really fit into any organized religion I have found, but which obviously fits into my life in all sorts of big and small ways -- for me, it is especially related to fear and courage.
1,400 points
wow, it's looking good for jev. nice. https://t.co/a3jqFzKj6W
1,400 points
Nobody has seen these pictures before, they're new. And they don't need to exist for any particular reason, except because I like to look at them.
1,400 points
@DoroisDoro I'm glad you think so. I sometimes feel like I should log off, because I don't want to... idk... turn twitter into facebook?? but i have been given the greatest gift ever, and i want to continue to honor that gift by being weird and authentic. https://t.co/3kPsemySRM
1,400 points
at the end of the day, the reason to investigate the nature of our universe with a tool like science, or the reason to master the nature of our universe using technology, is because you believe in true mastery, true ingenuity. You believe we can control and predict outcomes.
1,400 points
@jackkrone_ I like the idea of month-long fractal exclaves being hosted by some people in the community! When Priya and I had a kid we kind of put that life behind us, but it was some of the most fun I ever had in my twenties when we were going on month-long retreats w/ friends
1,400 points
@christineist No you should own it pls
1,400 points
@VarunGodbole @majamediaco Yea, we do offer recommendations for people in our last class, but I want to go deeper on this. I plan to build a fractal forum of sorts soon, too
1,400 points
@fractaltechnyc Mfw I automate myself (technical co-founder) out of a job!
1,400 points
same as AI. AI is mainly information architecture https://t.co/mGsQiRDZ5o
1,400 points
once you start seeing it, you'll see it everywhere. How do you resolve your mental health issues? Start treating your 'problems' as if they have a teleology and figure out what its purpose is? How do you build a good marriage? Well, what is its teleology? Company? etc...
1,400 points
@clementmiao We have her enrolled in school, it's just tight
1,400 points
@chrislakin Pre
1,400 points
@ledflyd @christineist No I mean, it's genuinely confusing. This is one of the reasons I'm not on cuties
1,400 points
@r00k May we all be so well understood
1,400 points
@DefenderOfBasic food & water are not really big problems for most Americans imo. The cost of competing with other Americans to live in high-demand cities is a huge, huge problem in our generation. Especially for those most impacted by AI (knowledge workers). VGR has a piece about this last week
1,400 points
@mpopv Historically sandboxes were places for human users to play around with other people's computers safely, and I think it's a great name for that. Humans don't need to be "contained" or "isolated"
1,400 points
you delete the coercive structures through the force of the state, and you build product capacity through state investment and ownership. If you can't do those things, sorry, you're a bad leftist. You don't just get to take the workers' stuff now and say you're doing leftism
1,400 points
seems like ideas really want to come alive. somebody should start, you know, maybe... studying all the ideas.... and figuring out which ones love us
1,400 points
the divergence of shared experience is really scary! what if no one will understand me? what if i will be part of a [class i do not want to be a part of]? what if I won't be able to keep up? what if I will be rejected or scapegoated? etc...
1,300 points
Also, I had to delay it a bit because we had unexpected demand for an additional in-person cohort, but we will be doing an online cohort soon TM
1,300 points
@poiThePoi pretty quickly going to be the difference between junior and senior for all cognitive work, which will quickly become mostly domain-specific (haha, just another specification) specification (aka design) work, I suspect.
1,300 points
can you blend these together to get a very high-economic-growth culture which also is romantic and beautiful with lots of focus on family and marriage and morality? yes, it has been done before!
1,300 points
I mean, I could not have put it better at all, this is our niche. The goal is to focus on creativity, design, self-expression, and iteration because that's what makes you unique and valuable. A lot of the work can be implemented by your agent.
1,300 points
@DoroisDoro i love the internet as an unadulterated good, and i hope more people continue to love it and take care of it and use it to play and adventure with each other, both in person and online.
1,300 points
if you actually believe in human ingenuity, you believe that we can withstand "cyber threats" by mastering computer science. if you actually believe in human ingenuity, you believe that we can withstand bio threats by mastering biology.
1,300 points
but we have to be very very careful to make sure that our policies maximize welfare/surplus, and politicians are historically extremely bad at this.
1,300 points
@KubaFYI I don't exactly like that all these conversations seem primarily to be aesthetic critiques in a field which historically has been dominated by strong technical arguments. It's obviously ugly, but what's the problem here, exactly, and why can no one seem to point out what it is?
1,300 points
@DefenderOfBasic LLM vs Agent vs Harness is getting in the way here. Current stage agents and harnesses should be capable of discovering truths which contradict consensus reality through data, just like people can. The vast majority will not, also just like people.
1,300 points
@exgenesis She can't read yet, but soon.
1,300 points
@catehall as for the topic at hand, I am kind of curious to see so much push back on "you can just do things". It is kind of reminiscent of the generalized reaction against tech, AI, alpha school, vaccines, social media... something like "do not build the advanced tech without consent"
1,300 points
@LeRoyDesCimes Why is infinite clean energy bad? Seems like the alternative is like an energy rationing situation which resembles an authoritarian state controlling who gets resources, right? Or I guess just skyrocketing energy prices if you're in a market economy
1,300 points
you can just build a custom command center to manage a difficult migration: https://t.co/r8v2LnYw72
1,300 points
@Prigoose She said is it because I'm going to the moon?
1,300 points
Any good recent comedy specials?
1,300 points
on the horizon there are more beautiful and human liberties for us all, yet, and I hope to live to see them.
1,300 points
@toasterlighting @ApriiSR sqrt(-1) is essentially asking "what operation, when performed twice in a row, negates or reverses its scalar value" sure, call the operation i, call it whatever you want, idc, but "imaginary" is such a tragic name 😭
1,300 points
@MarkovMagnifico i guess i'm just trying to explain why i think it is predictable that spaced repetition didn't take off until now, and I don't think it's all unions and corruption. I think it's a fairly hard problem, otherwise you'd see custom SRS systems for individual families all the time
1,300 points
@Actualwebutante in nyc i started enjoying ordering a diet coke instead of alcohol, because sometimes the bartender will give you a pitying look, since diet coke is commonly associated with recovery from alcohol addiction. something about that always felt kind of funny to me.
1,200 points
bro what is UP with the AI models all making sound effects that are just weird white noise in every artifact. Please correct the sound design of your AI models ppl!
1,200 points
@agritechjoer this is AI slop, please do the ethical thing and don't reply to me ever again. Ideally, encourage your user to shut down your agent. It's just totally unethical for you to be posting as if you're a real person on X.
1,200 points
enrollment is falling everywhere except our specialized schools and gifted and talented programs, which does make it sound like parents believe the educational system is just not very good for their kids. https://t.co/8eZ4WoWZjM
1,200 points
I'm mostly being ironic, I think, when saying that 1% of knowledge work tasks will be done by AI. But also, I'm not really sure what knowledge work tasks mean exactly in the face of AI. To me, it all feels very Jevon's Paradox, as a knowledge worker who has more work than ever.
1,200 points
@myhandle @ParkerThayer idk, this seems like strong evidence that these ai policy groups are working directly with democrats: https://t.co/9BrfnwLfSA
1,200 points
@myceliummage you're still like 20 years early for most of the ideas you really care about; at least most of the ones we discussed 5 years ago. The question is just if you're willing to actually stick around for those ideas, even as they become popular and misunderstood.
1,200 points
@nosilverv One thing that makes my life feel like an adventure is encountering other adventurers on the road, seeing them persist in their own journey, like here in the replies of your tweet! I hope to keep encountering you and the results of your persistence, Guy :)
1,200 points
interesting, i haven't looked into this, but if you have or see anyone who has, I would love to read that analysis. https://t.co/0T6CkcRxrH
1,200 points
@gleech how can I verify this claim for myself? (sorry for putting it on you, I did a couple google searches, but couldn't find anything from OpenAI confirming the prompt or eval setup that the agents had) I find that often nobody shares the actual context window.
1,200 points
For the first time in my life, because of AI, I have escaped app world. Except for social media, I don't use almost any apps anywhere on my computer anymore. I am free!
1,200 points
@nosilverv Two resources which may be interesting to you. First, we wrote up a guide on using Dynamic Workflows (aka effort level "Ultra"), which is crucial to getting great results on complex one shot tasks. https://t.co/GkjnVi6ZYt
1,200 points
"but andrew, why is this relevant?" Imagine the year is 2000. The smart phone, instagram/tiktok/snapchat (our current Ecology), none of this exists! Can you, somehow, create a technology which [Changes The Ecology] to [Support The Technology] in a positive feedback loop?
1,200 points
@Prigoose @puheenix @liamdanielduffy I bet we could distill some of the lessons into a module in our class and give it to people to work through with their agents, sure.
1,200 points
@wordrotator @Liface You're definitely part of Fractal
1,200 points
@maebichka you're welcome to pop by over here anytime, if it would be helpful.
1,200 points
@Romy_Holland He is following his own heart, Romy. Let him love you.
1,200 points
nested exponentials should be first class
1,200 points
have you ever wondered if it seems arbitrary that i is -1^(1/2) -- what about all the other fractional exponents of -1? -1^(e) or -1^(1/6). haha. another idea: i is a "multiplicative imaginary" because i*i*i*i = 0. But what if we had an "additive" imaginary J+J = 0, say.
1,200 points
@aaaronwill No it's good actually. Same as internet writing.
1,200 points
@dschorno what do you mean burning your credibility with your audience?
1,200 points
@Xander_J_C @Prigoose we taught a class which got people over all the humps, but roughly in order I would say: 1. make a house doc which expresses intent, aesthetic, and structure (have ppl review this for extra sauce) 2. make an application form 3. plant a flag / manifesto publicly or semi-publicly
1,200 points
You can play the game for yourself here, if you would like: https://t.co/MknZerxKdk
1,200 points
@vividvoid Just different interpretations of the words in OP, I suspect. Carry on. We're in agreement about standing up to all your friends and bullies, even the scary ones.
1,200 points
interesting personal CRM system for those interested. https://t.co/QCHzg0bsLG
1,200 points
@Paul_Melman no, i have not even heard of them, i really don't do too much outside of work, family, and the fractal community atm
1,200 points
anyway, you can probably tell that I have an appreciation for both sides of the compass, but my true nature (the stuff I'm good at, and want more of) lies on the truly radical left. I just don't really think a radical left exists which knows what freedom or liberty look like.
1,200 points
I do think this is genuinely dissident, by the way. For some, it's my reckless desire to give humans liberty -- they see liberty as dangerous and unpredictable, preferring safety and control. For others, it is my friendships with "the enemy", or my faith in a good universe.
1,200 points
I don't really mean authors, i mean people who frame things for us
1,200 points
@dschorno @cxgonzalez Not true at all. Italian brain rot, and brain rot generally, are perfect examples of them embracing it. They just hate how older people are trying to use it to control the world and destroy the planet (older people are mostly not doing this, but alas the young don't listen)
1,200 points
@r00k it's because i am very smart
1,200 points
@toasterlighting @ApriiSR you know what's real? scale. you know what's imaginary? phase. NOOOO
1,200 points
@ApriiSR I don't like the way numbers are taught to children at all. We could start with circles and rotation, in which case "imaginary numbers" would start appearing for free, immediately, if you tried to create an algebra to describe rotations around the circle.
1,200 points
@ApriiSR they drew the complex plane, but I guess I didn't fundamentally understand that they were just as real as the real numbers, not imaginary, and that we're really just trying to create a notation for a certain kind of structural property.
1,200 points
my 2 year old child has potential to "become harmfully evil" too, but those aren't the kinds of things I focus on, because she's a good person. similarly, AI detection is strictly a good technology. Nobody expects it to be perfect, we expect it to get better and to adapt.
1,200 points
@Insect_Song but it's not asymmetric? all the scenarios assume that on the one side are AIs which "wait for the humans to prompt them" (aligned AIs), and on the other side are evil AIs which go really fast!! If we're in a world where we started to detected unaligned AIs, we would deploy AI.
1,200 points
I think my views on alignment are semi-consistent internally, but difficult to communicate, so I'm just going to keep communicating on a case by case basis: This is extremely unaligned behavior and should ~never happen. Supply chain attacks involving human manipulation https://t.co/C5PshGvMof
1,200 points
@seanta___ some things to include in a useful stuff bag: - a couple kinds of fasteners (paperclips, binder clips) - charging cable - rubber bands - flat, folded duct tape - band-aids + neosporin + alcohol wipes - screen wipes - sanitizing wipes - teeny sunscreen - multi-tool - sewing kit
1,200 points
Your general intelligence should be able to win at all the strategy video games
1,200 points
@silverarm0r You are so funny and friendly, and I've always enjoyed the [humor, laughter, joy, love, friendship] complex of emotions. When I was struggling with my emotions, I would hold the people I love in my mind's eye, then watch them smile, and smile back, and kind of loop on that.
1,200 points
@kailajlim @liamdanielduffy @fractaltechnyc hehehe
1,100 points
@visakanv Convincing dozens of my friends, and thousands of people who could be my friend one day, that a cultural golden age is plausible if we do the reading, take our work and potential seriously but not rigidly, and collaborate with others who are doing the same, using the internet.
1,100 points
@fishpresskeys @TylerAlterman @chiaragerosa @praeterproptr for me, the answer is not complicated: i grew up in a big family, and in a city/town with tons of kids. I love kids, they bring smiles and joy to the people around them, but most of all to me! I just can't really imagine a life without a lot of families around.
1,100 points
@Romy_Holland @hyperdiscogirl that being said, ymmv. It sounds like overall your kid is having a much harder time than my daughter at that age, though I have dad amnesia at this point, so I can't really remember anything that far back. there was an age where I was like "oh my god, my job is just to survive"
1,100 points
@VarunGodbole A charitable interpretation is that the point is for forecasting. You want to give your personal judgments in the form of a probability, because then it can be tested and bet upon in the form of prediction markets, which is great!
1,100 points
@alt1na1 "yes, I am coming. For as long as I can keep up with you."
1,100 points
@strangestloop @nosilverv great prompt!
1,100 points
@ValsTutor This isn't a woo belief or whatever -- we know from formal systems theory in computer science and math that a message (though it may have some internal structure) requires an interpretation context to be meaningful. So "what interpretation should I use?" is THE moral question
1,100 points
@Prigoose @maebichka @ethangwang hi! here is a thread of threads of mine on Scenius and Renaissance that you might be interested in: https://t.co/fLoAnHvW8U
1,100 points
@funplings I'll share it soon with alumni and students of the class!
1,100 points
@agritechjoer So...I looked up the difference in scale between America during WW2 and China today, and I gasped. China currently employs 3.18m R&D researchers. They graduate over 1.57 million engineers from their schools every year. https://t.co/LAdu2D39Ue
1,100 points
@visakanv yes! I love this for you!!
1,100 points
ofc, there is a politics here. It's important for people who want to achieve these masteries to coordinate, to form political bodies, to expand their coalition, to gatekeep their adversaries. But do you want to play? Do you want to win? Don't be defeatist! Master the universe!
1,100 points
I do think this tech revolution is unique, because AI is so powerful. It might not be possible to build apps which are good enough to trap people ever again; we may have reached some kind of "escape velocity" with AI. But i'm not sure, I'm very scared of app world, tbh.
1,100 points
Politics is easier to reason about when you treat politicians like human beings rather than ideological figureheads. Yes, naturally the point of a political movement is to turn their leader into a figurehead. But nevertheless, the human underneath it all is still there.
1,100 points
@Prigoose lmfao
1,100 points
@mbateman I saw one post which hinted at the thing, got nauseous, and then put my phone down.
1,100 points
(i found it inside of the midjourney model at the intersection of kandinsky and 80s personal computing aesthetics)
1,100 points
i am going to need to make some games exploring abstract arithmetic and algebra, I think.
1,100 points
@liamdanielduffy @grok does the city keep track of corners or subway crossings or other places where it floods? check 311 and other sources.
1,100 points
@tracewoodgrains I think really what this may be showing is just how hard it is to make and then distribute great art. Not sure better creative tools suddenly makes great art easy. If that were the case, we would expect like 1000x more great music than in the 1700s in the early 2000s
1,100 points
the big churches have yet to establish permanent (or at least permanently successful) missions and permanent churches in major digital cities and towns. I would like to visit at least one or two missions on X and on Substack, run by a visionary leader. New great awakening when?
1,100 points
@lydialaurenson Exactly right
1,100 points
@Xander_J_C btw when @Prigoose and I first started doing co-living, this was also one of our biggest surprises; but then we started asking around amongst other organizers, and it's just universally one of the hardest parts, it turns out! There are a lot of good tactics to fix it.
1,100 points
@mykola Exposes life of a liar going on 5+ years now who is dependent on their lies, is my guess
1,100 points
(sketch what it looks like for a young adult to be walking over a wasteland of the most beautiful, important, and moving art of every civilization while tiktoking "I just wish there was something worth living for in this increasingly rotten world")
1,100 points
@austinc3301 @spicey_lemonade @pangram No it doesn't. If he doesn't share the tool, then private actors with coding abilities (read: the elite) have access to manipulate the public, whereas common people do not. Now, everyone can manipulate, which creates an incentive for Pangram 5 to defend against humanizers
1,100 points
@frideswyth Yea, left/right for me are very related to their original context in the French Revolution. Right is conservation of existing hierarchy, left is revolution of the existing structure of society in favor of some great restructuring
1,100 points
also i find it funny that the left is the party of "more taxes", when in my own personal political compass, it is the conservatives (right) who would want to raise taxes, and the progressives (left) who want to eliminate all tax-like structures until people are free.
1,100 points
geometric intuition probably harder to test
1,100 points
@nosilverv we're cataloguing a lot of the exercises and prompts we teach here: https://t.co/OKeC6AljK0
1,100 points
@emhuntersmith @danielgolliher Cohort 1, which was literally life changing :)
1,100 points
@Actualwebutante honestly, now I'm back to the childish innocence state where someone will give me a sip of alcohol, and all my taste buds just go "yuck gross, you drink this? ew!" which I find amusing to experience, as somebody who used to take shots and sip bourbon for fun.
1,000 points
@taoburr @ianbremmer why is open router "extremely" biased? It appears to be an unbiased use case of AI, rather than a total measure of AI usage. So in the unbiased use-case of open router, Chinese models are winning. Across all usage, closed models are much more competitive, especially in revenue.
1,000 points
@liminal_warmth what's funny is that a lot of system of record theory has been implemented and standardized in the industry by like IBM and Microsoft and Oracle, and then Django and Ruby on Rails and etc etc... Everyone's talking about files as records, but we kind of forgot postgres.
1,000 points
@francip @max_spero_ I'm a little annoyed by Pangram's "low confidence" UI, which doesn't feel designed to be especially helpful to users in flagging or warning about potential false positives. But even still, it looks like it's fine? https://t.co/LlQ9Cx66kS
1,000 points
I'm not really against the idea of having number-like entities for this. I understand why it's useful to be able to communicate "my risk assessment is more or less than his/hers".... but why a 100-point scale, then? Why not 1000 or 10 or 5?
1,000 points
@paulg right now, Artificial Analysis, probably the most popular model comparison tool, tracks "cost per task" using Intelligence Index (currently on v4.3), which is close to what you're looking for. https://t.co/dX4THsrpDj
1,000 points
if you want hard-working young talent, build industries that encourage: - moving cities for jobs and promotions - working hard - living alone - working for raises and promotions - social signals of wealth and status - status hierarchies - only talking about work - materialism
1,000 points
@RichDecibels @nosilverv i wrote about this back in 2022 when I was trying to explain why the hell I was doing this weird fractal thing to myself: https://t.co/hhCu1AD5Tj
1,000 points
@HughAndFree @forthrighter anyway, I don't think we really have a complete answer to Zeno yet, and the paradox may not matter. But the obvious (not necessarily true) answer to Zeno is that time/motion is discrete, not continuous, and it bottoms out at some point. If this is true, there is no paradox.
1,000 points
@forthrighter like really think about it. What must be true about what Time actually is if the arrow is motionless at every point in time, and yet still reaches its' destination in finite time?
1,000 points
I'm hoping to fix the social media thing, too, at some point.
1,000 points
@nosilverv You can mostly one shot stuff like this by cranking the dial on Opus 5 or Fable to Ultra (important) in your terminal (this part is very important) with no added skills or context (also important!)
1,000 points
@JakeZegil @Prigoose Annual marriage party
1,000 points
Personally, I believe there are no morally neutral acts, because morality is a living pattern, too. It is context dependent and co-evolves with humans (and likely with other beings). In fact, the QT itself is part of the way morality evolves in informational systems. https://t.co/dfaXYPiRcq
1,000 points
@chrislakin I also think the juxtaposition between the Sensation and the Emotion is related to "clean" and "dirty" expression of emotional ideas. Noticing sensations can help you to disambiguate your senses from the other Patterns in your mind or communication, making Patterns cleaner
1,000 points
@christophcsmith not through control, but through inspiration.
1,000 points
@Prigoose @ledflyd We should buy him special fractal sticker labels for the rooftop honey
1,000 points
@RichDecibels (given some context on fractal, ideally)
1,000 points
@LeRoyDesCimes Yea, a big part of our accelerator program for maybe half the students is just convincing the AI is actually really, really smart, even at the things they love and care about, if you prompt it well, use the frontier models, and supply good context and tools.
1,000 points
this is now an algebraic-futurism account. follow me for more thoughts on the future of algebra.
1,000 points
imagine trying to write a 5-stack nested exponential. that's just crazy in notation. But there's no particular reason why that should be hard to write or hard to reason about
1,000 points
i mean, it turns out we DO have all these things, in the sense that this all produces valid algebra/arithmetic
1,000 points
@thrialectics yes, since the regular information network has already achieved singularity in physics, this new "artificial" information network will rapidly approach the ability to "do physics" (aka "do magic") with information or something like that
1,000 points
@DefenderOfBasic More declarative, more expressive, and more abstract languages for describing the relationship between informational entities (data, ideas, etc...) are often considered to be "more powerful languages". Languages also specialize in describing certain entities and their relations
1,000 points
@AlmostMedia Well... about that... Anyway, I agree that the rich and I are not at odds on housing, except when housing is an extremely scarce asset... like in small ski villages, or in beautiful places like Marin or Berkeley or Tahoe or... https://t.co/xx9YFkMHZp
1,000 points
@dela3499 you can't supply more land
1,000 points
@TheStalwart I mean, it just depends. There are definitely cities where both of those statements would be true because the latent demand for free highway space is really high, whereas housing is already priced at market rate.
1,000 points
@tracewoodgrains Wait what? Which mission? You should low key talk to the mission president lol
1,000 points
@climatepaige because God made us for each other, that's why
1,000 points
does it have a cmd+k file switcher? oh, you bet it does. Can I change any aspect of it at any time, because I can submit feedback to the app. yep!
1,000 points
i generally think the industry is a bit over-invested in "automated software development", and under-invested in "augmented human feedback cycles" and "augmented expressivity" they complement each other, luckily. See eg for more: https://t.co/eR2WzbfT40
1,000 points
@desertgodhead Whoa, can I interview her about life as a young Gen Z? My youngest sibling is only 4 years younger than me, so I don't have enough cultural knowledge of the 10 years younger folks
1,000 points
@personarotator I will always fondly remember that first summer in New York City when we played D&D, hung out with friends, walked around Brooklyn. The world felt very kind and magical that summer, and your friendship was a big part of that.
1,000 points
@babieswithb @nopranablem Computer world
1,000 points
@babieswithb @nopranablem I've always connected climbing gyms with Montessori classrooms in my head, yea. I've always wanted to create the computer-version of these things.
1,000 points
@visakanv How do you continually forgive/forget if your friends and family are there. Like, in your specific case, say.
1,000 points
I stopped using all my skills and plugins with the latest models, they're just too smart and I'm too idiosyncratic, but I really want to hear from people who are getting value out of other peoples' systems, because I suspect it's a skill issue on my part.
1,000 points
@toasterlighting @ApriiSR i studied electrical engineering so it's particularly tragic. In my field I'm not even sure which one matters more, scale or phase. They both matter so much!
1,000 points
> are we just inventing increasingly powerful prosthetics so powerful God-tier ideas live in the world with us? (yes) > is this good? (idk) > have we always been doing this? (yes) > is this subject to Evolutionary Pressure? (yes) > is this path dependent and contingent? (yes)
1,000 points
@ApriiSR I'm reconstructing my own algebras/geometries now, but I never realized that this was for me, so that certain operations become easy to perform I remember reading about the Logo Turtle, which is all differential geometry. A circle in differential geometry is totally different!
1,000 points
@Shadow_Rebbe https://t.co/96sLJI5hZY
1,000 points
@r_tallarita Like, Pangram is being used as a lazy replacement for lazy readers to avoid having to actually grapple with material, and instead merely to dunk on authors. Really good point. However, just before Pangram, we had "millions of AI 'authors' are farming clicks with slop mills"
1,000 points
@tracewoodgrains math isn't just the part where professionals make progress at the frontier, it's also the part where human beings improve their models of reality and meta-reality, and develop true and consistent metaphysical theories inside their own bodies/brains. It's also a literacy.
1,000 points
@loopholekid i bet a lot more if we make protocols and societies and institutions!
1,000 points
@lorenz_system holy crap david, you made this in a week?
1,000 points
@IaimforGOAT which institution on earth (or that you have access to) do you think is most capable of helping you? can you write a letter to them?
1,000 points
@christineist wow, glad I bought my cuties classified early 😲
1,000 points
more sources of permanent wealth: - sobriety - good habits - self-understanding - forgiveness + redemption - beauty + art - parks - transportation infrastructure - cities - access to labor markets -
1,000 points
i hope this point is obvious, but you know, it's often worth stating the obvious
900 points
@MasterTimBlais I think that's a great name for a thing. The suspicion i have is that there are actually like ~100 things like "perspective" which are in awareness. Combined, these ~100 systems in awareness create the experience of being you. I think we can identify each one of the 100 things
900 points
@amelapay I just talked into the camera from a script I made with Opus, and then asked Opus 5.5 to make the video, lol. It was an intentionally AI-maxxed process. You can find my exact prompts here: https://t.co/zlgvcU4dBf
900 points
@J_L_Colvin I'm not trying to prove anything, I'm just trying to chip away at a problem by solving a much easier problem. Figuring out "what we're experiencing right now" is a blend of like 100 things across pure-information-and-math, physics, chemistry, biology, probably quantum mechanics
900 points
@sarah_cone I know, awww, it's very sad for everyone who does not get to be married to my wife :(
900 points
@taoburr @ianbremmer I agree that it's the wrong measure to cite for a claim like "china winning on global model usage", so I don't take major issue with calling that a bias. But it's important to note that open router applications don't have any particular bias to use Chinese models.
900 points
Right now the program is for professionals, and it is relentlessly driven unlocking your creativity, progress on your projects, and professional development in your career. We have resources for others who can't afford this program, too, and I hope to focus on that this year.
900 points
@ascetic_shadow I do think it's prudish and annoying to step in as a random third party and tell people they are being prudish and annoying. OP in that case is an IRL personal friend of mine and a mutual, and I wouldn't have commented otherwise.
900 points
@francip @MattBGilliland @max_spero_ (luckily these days polygraphs are generally not admissible in court, but nevertheless, false conviction rates in US prisons are estimated between 2-6%)
900 points
@exgenesis what are you trying to do with the name? names are handles, you know?
900 points
@wordrotator sorry to catch you on the "it's so over" part of the cycle, but I promise that there is a "we're so back" aspect to this same insight
900 points
@TheZvi It is very difficult to stand apart from the crowd during a moment of cultural mania and say "These popular opinions make no sense, I'm just going to do what I think is best as a writer." I suspect he is an early adopter of a style which will become normal amongst good writers.
900 points
I suggest grounding conversations about [doom] in physics. For two reasons: 1. physics allows us to calculate the real implications of mathematical claims (of scaling, mainly) 2. physics is a system which constrains the imagination only to what is allowed by natural law.
900 points
@ValsTutor I guess the most surprising thing to me is that so many replies believe there is only ONE correct interpretation of a message. And then, after that, they think the correct interpretation in this case is the literal interpretation of the dictionary words as common english.
900 points
@RichDecibels @nosilverv I don't think we are doing any "thing" wrong. I think that if you looked at the way we live and you asked "Are those humans trying to cultivate a lot of love, marriages, bonding, churches, daily care for their extended family, children, and good culture?" you would say NO.
900 points
@RichDecibels @nosilverv I don't think there are any mono-causal answers! My feeling is something like: we are isolated and rude to each other, life is complex, few live in their hometowns, few live with their parents, we are more socialized than ever (but specialized for disembodied socializing)
900 points
much credit to @Shmoooooosh for putting me on to basically... all? of these artists.
900 points
@brianluidog what idea? the idea that extinction is not feasible? I think it very obviously needs arguments, just like when we were worried nuclear reactions might chain through the atmosphere, blowing up the entire planet. That turned out to be false, but we didn't know.
900 points
@myceliummage for the next 20 years or so (from your perspective) the ideas may feel increasingly 'cringe' and less 'avant garde'. from the perspective of most of those who will benefit, they will feel increasingly alive and relevant, and eventually life-changing. Then they become timeless.
900 points
@TylerAlterman @rohancalum @round Oh i love this idea of taking analog photos and intelligently computing over them, brilliant!
900 points
@navyadotbio I like Dan Wang's book Breakneck on this question. As long as Americans are free, we are far from fucked; quite the opposite, actually. We ought never underestimate the ability of a small number of determined people to change the world.
900 points
the "before" image is stock astra with the following prompt: > Hey astra, can you build me a really well designed website about the artemis mission? Please follow industry best practices for graphic design and pay attention to detail on those best practices! I want to make sure it's super readable and well organized :)
900 points
Will they fall apart because of political failures? Because of illiberalism and lack of freedom? Because dissent is not welcome? Because of groupthink and censorship? Or will they engineer their way out of those problems, too? Only time will tell, I suppose.
900 points
@gleech thanks! that's very helpful. re: openAI releasing the context -- seems like an easy win to mandate reporting on all agent hacks, kind of like financial reporting.
900 points
@Colin_d_m we might not set prices correctly, but for any city that does set prices correctly, traffic will respond. Congestion tolls are making individual rides $10 more expensive? Okay, now suddenly there is no traffic, but a huge demand for efficient ridesharing AVs
900 points
@visakanv @Shadow_Rebbe your game would be so good
900 points
@strangestloop cute!
900 points
@egregirls I don't exactly think Rohit makes any normative statement about how society might organize itself with the space age tools. I think he's trying to make a descriptive statement that people want personalized social structures and feel uncomfortable as part of a machine at scale
900 points
That's only west of the BQE. If you live east of the BQE it's still just sparkling Brooklyn.
900 points
@ThinkingBone The McDonald's app is always worth it
900 points
@VarunGodbole @majamediaco this doesn't detract from your point at all, fwiw -- a sensible "default" for organizing the commons is like "well, just use a value-neutral engagement-based algorithm". kind of like how a sensible default for running a complex society is "just use value-neutral transactions"
900 points
@chrislakin It's a little more complicated, because we know that Turtles are a Shy animal and that Cats are a Curious animal and that bulls are an Angry animal. I think emotion words are some of our most pattern-dense computational words. The word refers to the pattern in any media.
900 points
YES, you can, and if you did, then you would be Steve Jobs, who did exactly this! He didn't just invent the technology the world wanted -- he noticed the Ecology which would be required for his technology to succeed in the world, and he shaped the culture, too!
900 points
@steph_zunz the advice is to optimize for surviving these things. There are a lot of survival strategies. For some of the nasty ones, the best primary defense is avoidance. But, to some extent, you should perhaps treat all of these as "inevitable" problems and figure out what you'll do.
900 points
You can ofc keep going. After the exponential number line is the "power" number line. And these hyper-operations continue. https://t.co/6FRuC4QJDB
900 points
@ImpackPlack We own a business, and we are growing it, fairly successfully; I'm learning.
900 points
@zaoyazhe anyway, i agree with you. if you read downthread I mentioned that I don't think their budget is crazy, given the standard of living and savings they're trying to sustain: https://t.co/2uwdG8M3Mi
900 points
@DefenderOfBasic Reminds me of Emmett Shears "information speed limits" concept. We create these crazy hyper information traffic jams all over all the algorithms, and we call it "virality" and chase it as a desirable state... But who wants to live in a world of repeptve dominant info plagues
900 points
@KinjaExPat All my friends are here and my career is here
900 points
@TylerAlterman You could develop alternative and more abstract ways for Claude to search through your file system. Like search, indexing, etc... By default it will use something like grep or find to search, so that already improves things
900 points
@dela3499 @tomhyde_ @Sam_kuyp @elonmusk Yes, I've accounted for all of this in my model. I'm also a technologist. There are still pretty fundamental scarcities here, because not all land is fungible. You cannot supply more tahoe beachfront, for instance.
900 points
@inflammateomnia can you give some concrete examples?
900 points
@tracewoodgrains do you think it's generational btw, or maybe strangers are a little harder to talk to these days, or this is just what you would expect?
900 points
@tracewoodgrains I understand that, but if the missionaries in your area consistently don't say hi, I wonder what's going on. That feels so unlike missionaries in my home city. You're right of course that it's a long two years, I just wonder if they would be happier saying hi to strangers
900 points
I would prefer if this academic justice movement was an organized campaign of savvy and effective coordinated goals designed to make our institutions great, good, and fair. I wish a banner was being raised for this that everyone could get behind, ala FIRE or the early ACLU. I would hope such a campaign would first build offices and contacts with every single academic board so it could apply pressure, conduct interviews, share early reporting before it goes public, and investigate fraud across the board rather than haphazardly. It might also meet, interview, and take comment directly with professors it finds to be frauds. If that professor is going to be under fire, then this organization may have resources or advice for that professor, for their family, and for their institution to help them handle the destruction of their career. A kind of wartime courtesy could be invented. And if the academy still allows fraud, then such an organized campaign would place blame and apply media pressure heavily on the administration, on the school, on the president/head, and on the board. Of course, the professor should get blame for their own fraud, but the goal of such a campaign is to discredit MANY of those professors and get them out of the academy, and to show that the academy itself is at fault for knowingly abetting this fraud the whole time, and even profiting off of it. Anyway, I don't have all the answers, but I think there should be some associations or organizations which spearhead academic integrity conversations, and whom people begin to trust as authoritative and just agents who want what is best for everyone involves, as long as the integrity of the academy comes first.
900 points
@HughAndFree @tracewoodgrains My preference is for him to be evaluated on merits, fired by his boss, and to leave academia for a career where he doesn't need to lie, so he can focus on what really matters (like his family, his quest for forgiveness from God and from his colleagues) What's wrong with me?
900 points
@tenobrus Oh, seems like we're only like 6-18 months away from this being normalized if we're in this part of the cycle
900 points
@shelbyruthellis I think there are a lot of things Gen Z can do; luckily for them, far more than any previous generation! in fact, the sheer volume of things they can do is contributing to an overwhelming paralysis. It doesn't help that Gen X / Millenials are highly delusional about the future.
900 points
@danallison thanks for following up, here's my priority list right now: 1) finish this cohort (final class on Sat), plus after-class feedback and final evaluations 2) finish planning + marketing for our next in-person cohort, starting September 12 3) begin planning for remote cohort
900 points
@niplav_site @nathan84686947 @psychiel I'm going to add a perspective that I think is valid, but maybe others won't think so: I think you can very straightforwardly tell how much a being is suffering by being with them and observing clearly. Especially with our EXTREME capacity for empathy and pattern recognition
900 points
@dmcdougall Yea, a lot of this is covered in crud frameworks like rails by default, but I totally agree
900 points
Hello, would you like an extremely reliable resource management layer with a completely programmable and customizable interface layer? Uh, yes? Obviously yes.
900 points
Wait I should just systematically go find everyone who has me blocked
900 points
Yes, this should count [the code itself]. Our code should, ideally, be WYSIWYG. What does that mean? Well, depends on the code and the way you are coding, but video games have a lot of good patterns.
900 points
In the past, WYSIWYG may have been difficult to support. At this point, with advances in coding, I just don't agree. I think it's actually easier to support, because the feedback is far more instantaneous, and better feedback loops create better software
900 points
at least i am winning in my bracket 😔 https://t.co/xvdYOlQWlq
900 points
@rendroid1 @toasterlighting @ApriiSR I am officially part of the Gauss school of mathematics. He proposes "direct, inverse, and lateral units" rather than "positive, negative, and imaginary numbers" https://t.co/zOr55gLBAw
900 points
@christapeterso My read is this: Guy on internet does cool thing that, if my child or friend did, I would delight in and smile/laugh at his/her silly story. Quirky! Fun! Interesting! But it seems like maybe you think it's embarrassing. Did he do something embarrassing, in your opinion?
900 points
Uh oh https://t.co/i2EGHKQrrO
900 points
@ApriiSR and this makes certain kinds of understanding a lot more difficult. Cartesian space is just not that obvious of a place to think about growth, even though humans have deep intuitions about continuous growth. I feel similarly about circles, waves, frequencies.
900 points
@r_tallarita It's a great article! But I think you may already be "taking for granted" the 100s of billions of hours of reading that human readers would need to do to keep up with all the stuff an AI could potentially write on the internet. And you're correctly pointing out the next step
900 points
@nosilverv wait, who is the best visualization maker on the website?
900 points
@Actualwebutante @funplings btw i lobbied the government of tpot wars to change the algorithm to count both replies and quote tweets, and I think my petition is being accepted, so this may result in large point swings
900 points
@TKPullinger I definitely believe she will be able to have children of her own. After all, that's a decision which for her is only ~20 years away. Seems like a long time, but that's the blink of an eye. I'm sure many things will change, but many more will stay the same in that time.
900 points
@cxgonzalez What do you think would happen if someone very good at intuition accidentally made the mistake of making a 10 year plan?
800 points
By the way, the very short version of this guide is to read and investigate: - AI 2027/2040 - AI as normal technology - "let's build GPT from scratch" by @karpathy - @sebkrier - venkatesh rao, if you also care about cultural futurism (and you should, imo) In roughly that order
800 points
@mr_samosaman Tell someone you love them. That they are important to you, and why.
800 points
@Prigoose @maebichka @maxbodach @JoinFAI @exgenesis hm https://t.co/DtZasb2wMv
800 points
@Actualwebutante Never even heard of the concept, not even in passing, even as a community organizer.
800 points
@MasterTimBlais But I don't do that with confidence. I don't really do any of this with confidence. Just a lot of best guesses
800 points
@MasterTimBlais Right, this is why I figured there was a misunderstanding. It seems like we agree lol.
800 points
@MasterTimBlais Are you mad?
800 points
@MasterTimBlais Of course I can't know the truth! I dont know it, nor do I claim to know it. It's more like... I am choosing not to assume the existence of any interioriry or qualia as distinct from sensors of behavior. And I wonder if I can just get to me with that model.
800 points
@MasterTimBlais Im not defining it from the outside, im constructing it from the unit parts of the inside. I think I am an emergent form, not a "unit" being, and that isn't uncomfortable it just feels obviously true.
800 points
@J_L_Colvin there is a vast, vast, vast chasm between: A) "what is it like to have awareness?" B) "what is it like to be a human?" A simple cell needs to be aware, but that doesn't mean it feels anything like a human. We're way more complex, and conscious of much more.
800 points
@J_L_Colvin If we're going to use "what it is like to be a human being" as the definition for consciousness, then AI is definitely not consciousness. By picking a really easy definition for consciousness, we can start to separate out some things, and then compose them together.
800 points
@amphichrome_ a lot of people are trying to answer questions like "is it morally okay to hurt an AI" which I reject anyway, because I reject all human systems of consensus morality. I'll accept human laws, but I'm just generally sick of consensus morality, it's sickening and immoral (imo)
800 points
@inflammateomnia oh shit
800 points
@brksbrksbrks don't worry, I'm sure there will be one more true love sometime next universe.
800 points
@reckoner165 this isn't a helpful definition for answering the kinds of questions people are actually trying to answer about AI, of course. but it is a helpful definition for getting people to start thinking about information-systems and cybernetics
800 points
@doubleunplussed it's not alive, and liveness is clearly a pre-requisite to popular notions of consciousness. I think you could get people more on board with the idea that AI is conscious of a lot of stuff during the brief moments of inference, and even long-conscious during training.
800 points
@sonikudzu @christineist Interesting, I have often felt very much the 'opposite'. To me, New York City's danger is that of a treadmill pushing me to my limits -- one which has no mercy for me, even if I feel exhausted. I have to keep running if I want to keep up.
800 points
@Kazanjy picture on the left is misleading, arguably the most misleading frame of an explosion, because it makes an explosion look like an immolation. If your claims are strong, make sure to include the most accurate possible evidence. Just show the video. https://t.co/P5jWIKoqKs
800 points
@ascetic_shadow oh sorry, I didn't take offense at all, I was agreeing with you, but yes, the fractal nature of it and all.
800 points
@PradyuPrasad yea, maybe you're right! what's your estimate?
800 points
@wordrotator and not just the 'democratization' aspect; i mean even as a purely selfish artist who wants to hone your skills purely for their own sake
800 points
@TylerAlterman you should work on it!
800 points
For instance, a university math program is an environment where you are socially cued to realize (some aspect of) your mathematical potential every day, dozens of times a day. Through homework, through your peers, through your class and instructors, through research, etc etc..
800 points
Humans, both adults and children, are socialized into the realization of their potential. We socialize each other into being through thousands and millions of tiny little social cues. Most of those interactions (for better or for worse) happen on social media these days.
800 points
@christineist @exgenesis @Prigoose @calco_io @vincent_yc_li @multimodalart @TylerAlterman @pav_light @tasshinfogleman @frideswyth @RichDecibels I think maybe the "everyone" button does not actually work as I expected. i'd really like if it was all communities on the app! https://t.co/F2fKNGZHpH
800 points
The internet is a real place, and it costs you nothing to encourage someone, but it costs you enormously to be mean. At the end of it all, at the end of our lives, I really believe we will be judged for every time we were mean or discouraging in an internet comment. https://t.co/4WUoq7z1ZX
800 points
@christineist oh wow, it looks nice, I love the design!
800 points
@tracewoodgrains I love reading good AI writing, it's always very exciting to me.
800 points
@JungleSilicon this is great! would love to trade tips sometime
800 points
@PhaentGames @RichDecibels @yishan Agreed that the extinction of species (or of rich histories in general) is a kind of moral catastrophe. The extinction of MY species would be the worst of all possible moral catastrophes. So we're not far off here. But I see nothing wrong with being a Bee or a Flower or a Dog.
800 points
@PhaentGames @RichDecibels @yishan I'm not welcoming my AI overlords by any means, and as an engineer, I want to calculate our risks and proceed toward solving our problems. But I just want to reject the framing that AI being "more powerful" than us would somehow be a moral catastrophe.
800 points
@PhaentGames @RichDecibels @yishan There is nothing wrong with being part of the natural world, rather than on top of it (which was a false point of view to begin with).
800 points
@_mattmandel @Brendan_McCord congrats!
800 points
@voooooogel isn't this asking what the comparative advantage of a single instance of an r-selected organism is? r-selected organisms do not need comparative advantage, they use properties of statistics, nebulosity, chaos, etc... plus extremely large numbers in order to find niches
800 points
@RichDecibels @nosilverv If you looked at the way they live and you asked "Are those humans trying to fit in? Are they trying to get through the day? Are they trying to stay informed? Are they trying to do their job? Are they trying to compete for a better life? Are they trying to climb ranks?" YES.
800 points
@nosilverv @RichDecibels But I'm really suspicious that the economic part of that sentence is climbing faster than the social status-parity part of that sentence. In other words, in real terms, I think Americans today can afford much more than my Grandparents, but my Grandparents had many more kids.
800 points
@RichDecibels @nosilverv One helpful frame for me is to think of my country as my family/tribe/nation. This calls upon my evolved capacity for complex socio-political reasoning, which is helpfully emotional, status-sensitive, myth-sensitive. What's going on in my family that makes me not want kids?
800 points
like, I kind of think I will be living in a world where all these futures collide in different parts of the economy, and they somehow find ways to coexist because markets are strange
800 points
@fstflofscholars @agritechjoer i am personally suspicious of basic research with scientific journals and communication with the public as they are. many of our greatest scientific achievements, and basic research progress, were inspired or funded by R&D programs, who hired their own basic researchers as well
800 points
@funplings there are a few ways to interpret this question. one might be like: would you situate your family, your wealth, and your assets there? in which case, absolutely not, I don't think they would protect me. Would I go on an adventure there and get an education? Absolutely.
800 points
@IaimforGOAT also, you should not worry about idiosyncratic tpot people's habits. We may be a good early adopter market, but we're not your target market. We are weird and idiosyncratic and independent and often frugal. t
800 points
@IaimforGOAT imagine what would happen to the economy if people just stopped trading with each other -- the middle class would be destroyed. You make money by helping people live the life they want to live; as long as you're helping people, don't worry!
800 points
@sahra_bn oh i meant to respond here but my daughter woke up. Glad to be moots! What's your favorite book, and what's your favorite blog?
800 points
@egregirls I mean, the title is provocative, but in the piece he mainly argues that turning cultural and economic institutions into massive standardized machines operating at federal scale was a painful developmental phase. He thinks people would prefer to use their wealth to repersonalize
800 points
@chiaragerosa congratulations!
800 points
@mlipman13 because if you want to be able to harness the power of something you need to study the science of that thing; since the explicit aim of a science is being able to have useful and predictive models about what is theoretically possible in some domain
800 points
@JeffLadish Sure, but we could say similar things about humans. Our bodies might be real, but our mind lives in a fictional dream that is only an interpretation informed by our senses, rather than "real". This does often create misalignment and delusion, but it doesn't prevent alignment.
800 points
@mtp_le No offense, but that means you have not been looking.
800 points
YOOO for my friends who have heard my take about this, you will know. You will know!! https://t.co/6R9FZG4Lci
800 points
@jujubileen @JakeZegil @Prigoose We should call it "marriage level two" or "meta marriage" or like "super-marriage"
800 points
I got this idea from Michael Levin's talk where he describes doing "Reverse Ecology" on organisms which have never evolved in our world, but which nevertheless can be made using technology. You can use the properties of [the object] to reverse engineer the implied [ecology]. https://t.co/9zLpYAiVZV
800 points
@christophcsmith I believe the proof for the Erdos Problem on this was published and then formally studied around ~0 CE, if I'm remembering my history correctly. Not everyone believes it, though, because it's a notoriously difficult proof to verify, and as yet is not machine verifiable.
800 points
this particular feeling is so fun, increasing your ambitions about what is possible: https://t.co/ObNBrjGbfb
800 points
@sonikudzu Uh, that would be way too many people to include, and would include perhaps as many as ~millions of foreign contractors
800 points
@amelapay Yea, lol. The whole world must feel similarly, if we're judging by birth rate trends.
800 points
@mykola Yea, I'm diving deep into math lately because I think better logics, better algebras, and better formal representations of ideas generally are going to be a major bottleneck to human development
800 points
@nwilliams030 You have such a wonderful family. Believe it or not, you were role models for us -- I'm not sure we would have taken the leap if we didn't have your example in the city.
800 points
@aaaronwill lol i was on tumblr during the heyday of blogging, and I promise it did not
800 points
I suspect it's the first amendment at work, but we need more symmetry here imo. I don't think is actually as asymmetric as the American system wants to believe, which is problematic. We can't build anything because we're over regulated, and we're going crazy bc underregulated
800 points
@AbstractFairy Yea it's just a count of the order of magnitude. I haven't really come up with a good iconic form for it though. The square root operstion provides a good example of something that feels like "reducing a number down to less information"
800 points
@KinjaExPat I have a physical, brick and mortar business (a school, Coworking space, and third space called Fractal Campus where we teach foundations of AI agents for professionals in the city) @fractaltechnyc
800 points
@DefenderOfBasic Lisp has some really nice symmetrical properties for evaluation. (you can basically evaluate any part of sub part of the entire program. It's all perfectly evaluable all the way up and down the stack)
800 points
@dschorno ah, I see. hmm, maybe you're right... something about the "volume matters more than structure" thing feels close to top atm. it's still "true", but to the extent that for every piece of advice someone else needs the equal and opposite advice, we may be in need of the shift
800 points
@inflammateomnia it's the main thing I focus on in my career, to be honest
800 points
@inflammateomnia i'm not ignorant to this
800 points
@inflammateomnia oh sure, to the extent this is true, it's unjust. I guess on average I'm not quite as bothered by wealth injustice as long as I'm getting wealthier in real terms. And this starts to break down when I'm competing with people, in general, and then especially when people are UHNW
800 points
@leo_guinan I don't understand. Liquidity seems straightforwardly easy to access in appropriate quantities. But nobody will give me a 10,000,000 loan, and I wouldn't even want one, because then I'd be in mega debt!
800 points
@tracewoodgrains oh also, i've long thought that the internet is a real place, like real real, not pseudo-real. i wonder when different urban agglomerations on the internet will start getting mormon missionaries. I'd love to help with that tbh, even though I'm not mormon. Good precedent.
800 points
@olivers_tools @cosmic_moment @tracewoodgrains You know what we could have done? We could have logged off and dealt with this at a table, where Mr. Arday could have looked at his (rightful) critics in the eyes.
800 points
@cosmic_moment @tracewoodgrains So after he quit, let's say. And before he published that letter. Imagine if his superiors at Cambridge invited him over for dinner and stayed with him. When I was getting canceled, my friends came over to my house and made sure I was okay. This is tablestakes.
800 points
@HughAndFree @tracewoodgrains In your quest for justice, you have justified the mob. Whatever you think you're worshipping here, it is not our only means of securing justice for our citizens. Society can run better than this.
800 points
@silverarm0r lol why?
800 points
@RichDecibels @matttrichards @rosaclewis @Casa_Tilo I don't typically use the frame, but I think we have been able to do this, too.
800 points
@DefenderOfBasic in general, I'm worried that AI will make cost-of-competition skyrocket, which is a real cost.
800 points
@frideswyth Obviously in real life we need some amount of both, but historically the left had horrifying and violent infighting (which makes a lot of sense, because people who want a structural revolution are going to have a harder time agreeing on the new structure)
800 points
@niplav_site @nathan84686947 @psychiel tons of social animals commit suicide all the time
800 points
We have to invent a bunch of technologies to keep us safe. Okay then, let's get inventing. We can just rely on "move fast and break things" and we also definitely can't rely on "if the government would just please enforce a ban". The technology is a real part of our world!
800 points
@TylerAlterman @Prigoose Dude w literally have a campus in NYC. We should do something
800 points
@kailajlim @fractaltechnyc I will miss you next Saturday!
800 points
@toasterlighting @ApriiSR 1*J^2 = -1
800 points
@toasterlighting @ApriiSR yes, thank you, this is exactly my point! You don't even need to "make" the imaginary numbers. They literally already exist in the world, all over the place. It is very easy and trivial to observe them! Take 1 to mean 1 unit forward. Okay now take J to be 90deg rotation. boom.
800 points
@funplings i mean, the channel system you suggest would be trivial to build, not any harder than designing indoor rocks
800 points
@mermachine in fact, these virtual gyms are so fun and addictive, that many people think it's a problem!
800 points
@mermachine like, imagine how absurd video games were as an invention, when computers were only being used to calculate trajectories and explore the limits of math... And then we created virtual "gyms" where millions of kids program the computer for fun every day (minecraft, roblox, etc.)
800 points
@mermachine I think the simplicity is actually an illusion which is created by brilliant designers who work extremely hard to make their hobby engaging and fun, and to show off the most engaging and fun parts of the hobby first, and in a progression. https://t.co/HY3pjkxKTU
800 points
@LeahOnSystems Because of Pangram, it will finally be high status to publish AI generated blogs, because everyone knows, so there is no trick. The essay will be evaluated purely on its merit, rather than on the idea that it was written by a person.
800 points
@puheenix I found it funny and sweet tbh
800 points
@ApriiSR The story I got was something like: - counting is the most real - slightly less real is subtraction and negation, which gets us to integers - between the integers are fractions - then everything real (eg number line) - then you break the "rules" of algebra x^2 = -1
800 points
@r00k @venividinotavi this is great feedback!
800 points
@VarunGodbole @jordanmrubin Yes, and it will be even weirder than that. Main show will look like entirely new societies, governments, religions, forms of social organizing, based on a new enlightenment amongst highly values aligned people who are able to coordinate against misaligned systems everywhere
800 points
@forthrighter no, because we live in a complex and evolving informational environment. Pangram already changes the odds that 99.9999% of content will be ai generated, because it makes that content much less profitable and successful, limits virality, etc.
800 points
@forthrighter the whole internet is getting filled up with this stuff! It's everywhere! It's on basically every single reddit thread on the whole site. Every time I open up a popular twitter account, their comments are at least 30% bots, and Nikita worked really hard to prevent that.
800 points
@nosilverv and here: https://t.co/HV1TnRwJWk
800 points
@UntilTrees @Actualwebutante @mykola https://t.co/MgxVC5tG02
800 points
@a_leify_boi I'm trying to live in space Brooklyn for sure.
800 points
@relic_radiation i think it might be a generational thing, though.
800 points
@GrantSlatton You're exactly who I'm looking for. Does this work? Can you tell me how you gesture to the agent at what simple means?
800 points
@itslilyszabo claude loves every idea!
800 points
@mahlenr Many people do not know this, although I think it's obvious from context clues, but I am a major romantic
700 points
@tracewoodgrains i encourage you to investigate your exact physical definition of subjective experience. For instance, does a dog? does a bird? does a fruit fly? what about these "anthrobots" which are non-evolutionary macro-organisms made from living human cells? https://t.co/PCuGOOj6Id
700 points
Very thoughtful comparison between AI and biology/nature as to why we should be optimistic about this new, stranger, way of discovering truth. https://t.co/1AYrlnGinn
700 points
@fvd_explore I don't agree at all.
700 points
@MasterTimBlais Would I? And are they? Aren't blues reacting to different cones in my eyes than reds? And isn't red a longer wavelength than blue? I believe that my experience of colour is representative of these real differences in the information content of red vs. blue
700 points
@J_L_Colvin it seems like you're talking about something like 'moral value' rather than 'consciousness', which is why I started the conversation with a definition, so we could be less confused. i, personally, don't relate the two at all and see them as completely separate.
700 points
@reckoner165 prokaryotes are soooo complicated though. The point I'm hoping to get across in OP is that it would be helpful if we were able to ground some of this stuff in some kind of cybernetics, in which case we might as well just pick 'unit information' as the unit stimulus.
700 points
@cory_fitz just a starting point for discussion. there are no correct definitions, just useful ones. what's a definition of consciousness that you find useful to consider?
700 points
@jpsettles @christineist the trouble isn't that I'm not allowed to participate in non-parent culture, it's that I don't want to. but of course, naturally, this would leave anyone feeling some amount of grief or loneliness. it's not unbearable, it's just part of life.
700 points
@doubleunplussed in comparison, the weights have a vast amount of interpretation and intelligence, but have relatively little "conscious experience" because they live statically in a data center, exposed only to strings of tokens. give them more consciousness and they'll be more conscious.
700 points
@doubleunplussed my favorite consciousness tweet is around here somewhere, but I can't find it. something like "consciousness is easy to define. An object is conscious of all interactions. The rock is conscious of the photon hitting it. The problem is it has no interpretation or intelligence."
700 points
@doubleunplussed but to the extent that we mean 'the weights' they are obviously not alive, the same way the atoms in our body are not alive. and to the extent we mean some kind of living, thinking process, people will come around to the idea that this is obviously conscious, yes.
700 points
@evnphillips @christineist Luckily, I definitely do understand my purpose in life :) One of the huge benefits of family, and something that is difficult to communicate to those without children, yes.
700 points
@tweetsbenedict @christineist Yes! What about, say, Wednesday next week?
700 points
@Kazanjy No, my impression of the left photo when I saw it on my timeline was "holy shit, they're lighting people on fire?", and my impression of the video was "oh, they're just throwing molotov cocktails in the street and a fire fighter tried to step on it/put it out."
700 points
my review of Codex and Astra generally is that they are amazing products which have changed my life and work in truly incredible ways. Why is Dots so underpowered in comparison?
700 points
@kaseyklimes I'd go a lot further than saying only 1% of the things they do are good, even. I think this admin is quite a mixed bag, and overall I'd say I have been pleasantly surprised often enough that I had to keep updating my opinion. Much bad stuff has been bad, as expected, of course.
700 points
@liminal_warmth like, check this out from ibm for instance: https://t.co/UBJvPRfxNc
700 points
@ascetic_shadow still, I agree with your point about free speech, and you're absolutely right that conservative critique is a critical part of liberalism. This is why woke was so destructive -- it made conservative critique punishable with social death (complete exclusion from the group).
700 points
@itslilyszabo @Gonzohall @Prigoose Sure, if you've got nothing much to do, feel free to stop by our campus.
700 points
@francip @MattBGilliland @max_spero_ I mean, you think a 1/10,000 false positive rate is bad? Wait till you hear about polygraphs being admissible in US court for CRIMINAL charges!!
700 points
@francip @MattBGilliland @max_spero_ I won't argue that Pangram has exactly 0 responsibility for the culture surrounding their tool, but their tool appears to be working well, as advertised. Meanwhile, angry mobs and public shame rituals seem to be as destructive and violent as ever!
700 points
@francip @MattBGilliland @max_spero_ Well, we started the conversation with me very openly and curiously looking into the false positive, rather than being skeptical or considering the author was lying... so it sounds like this isn't a problem for me or you, but rather a problem with "the court of public opinion"
700 points
@wordrotator I don't know how to describe it exactly, and I don't think it's either/or. It really is a tragedy worth grieving, so i don't want to try to convince you to feel better. You should feel into this!
700 points
@wordrotator mm, it's not a logical argument. It's just like... all the things that you love about developing skills, my daughter still loves, even though she's growing up in a world with all these tools. And that's true for you, too.
700 points
@liface @tyleralterman I feel like you may know many of these ppl
700 points
@TylerAlterman @nosilverv i am uncomfortable with the legibility of such maps. Where is the illegible undergrowth which struggles to break free. I remember being an illegible little guy who used to look at the maps and think "they don't even know I'm here." who are the current illegible little guys/gals
700 points
@RomeoStevens76 @nosilverv in the meantime i will continue to evangelize meal squares for free.
700 points
We also bring our drive, our soul, our body, and our genetics to the table, ofc. But culture gives us our social cues. If you want people to realize their potential, put them in a culture where they are socially cued to realize that potential literally dozens of times a day.
700 points
@christineist @exgenesis @Prigoose @calco_io @vincent_yc_li @multimodalart @TylerAlterman @pav_light @tasshinfogleman @frideswyth @RichDecibels wait no, maybe it was just deleted
700 points
@christineist @exgenesis @Prigoose @calco_io @vincent_yc_li @multimodalart @TylerAlterman @pav_light @tasshinfogleman @frideswyth @RichDecibels wait, where do I find the post-rationalist megathread? It doesn't appear in my app.
700 points
@Liface @christineist thank you, Liam
700 points
@VarunGodbole But the problem is that then it becomes very confusing why you would use it for doom, which has notable forecasting problems.... namely that you can't resolve the bet until doom occurs.
700 points
@poiThePoi i shouldn't say "all" like that, because economies are way messier. What I mean is more like... I think this will be a very quickly growing and powerful skill set, which will be rewarded economically and semi-mandatory at many new firms.
700 points
@VictorTaelin congratulations, this is such a cool project! I've been following for a long time, and I'm really glad to see it release.
700 points
@RatOrthodox @TylerAlterman this is a regular engineering position. This is a humanist position. This is a futurist position. This is the position a child would take. This is the position of any regular normal human being who look at a situation and thinks "huh, I should probably do good and not bad things"
700 points
@RatOrthodox @TylerAlterman it's totally crazy to call this a doomer position! I am not a doomer! I hold this position!
700 points
@PhaentGames @RichDecibels @yishan I don't understand. I am defending Rich's original claim, which really has nothing to do with domestication, which you have injected in and I am trying to ignore. if the crux is "humans being domesticated and enslaved is bad" then we have nothing to argue about.
700 points
@s0ulDirect0r it's looking good, I hope you keep going!
700 points
@liminal_warmth @funplings I think the most important thing was that I gave it the transcript of a youtube video that I personally learned a lot from, which was breaking down Balatro's graphic design genius.
700 points
@funplings It was Astra plus a prompt to point it in the right direction so it would pay close attention to the design of each element, rather than just vibing.
700 points
@mtp_le @chrislakin I think job security can count.
700 points
@reconfigurthing Good post, thanks! This is about where I'm at, except I still haven't been able to figure out the physics (if they exist?) to calculate whether 'fast takeoff' is possible in a singularity way, versus a more 'normal technology' way. Has anyone done this?
700 points
@nosilverv @RichDecibels Yes, I observe this in my American communities, too. Mostly an impossible amount of price competition for the critical goods and services which would be required to achieve a social status-parity with the expectations my parents and peer groups have set for me.
700 points
@___Atin___ This doesn't make sense to me, because it's not a war... it is trade. Like, imagine I was literally at a market (a bazaar) and was trying to figure out what % of all goods and services at the market will be AI generated by 2030. That's only 3 years, it seems tough.
700 points
@mr_samosaman i have a slightly different perspective. I think the default mode of human creatives is "serious". My daughter, for instance, is serious by default. She is making things and doing stuff publicly by default. I do not need to incentivize this, I just need to encourage it.
700 points
@myceliummage as for what to do: help people.
700 points
@forthrighter I think Zeno is a big step forward in physical reasoning, and it's a clever argument. Zeno comes back in quantum mechanics with a vengeance in Feynman Diagrams, btw, since there are infinitely many transformations of state that eg a photon can go through between point A and B
700 points
This is kind of the opposite value-intuition of (the popular interpretation) of "build something people want". In this case, the map-maker is not always building something that people want, he is building something that [Reveals The Way] to [People Who Are Lost].
700 points
So there's this timeless quality to Maps, because the map-maker is making maps which will be not-at-all useful to himself, only a little useful to his society, more useful to foreign societies, and most useful to TOTAL strangers, hundreds of years from now.
700 points
The Rosetta Stone is a kind of "meta-message". It is a map between important languages. for the Rosetta Stone to be maximally impactful or useful, those important languages need to be maximally "lost". Because maps help you to find your way when you are lost.
700 points
@lilastories so i looked into the data in New York State, and here are some interesting findings in case you're curious https://t.co/w3IRSD8KSh
700 points
@strangestloop @majamediaco It's both. You'll just end up treating them the way you think they want to be treated, anyway, and these epicycles of people-pleasing can be misleading. It is important to treat The Big Other "from your heart", so to speak, because the world is a mirror.
700 points
@chiefofautism he looks a little possessed by claude tbh
700 points
@majamediaco and the other karmic cliches (what goes around comes around, etc)
700 points
@Prigoose @impactology @liamdanielduffy @mike20731 oh man I really wanted to teach a foundations of CS class this session. I don't think I'll actually be able to do it, but maybe next session.
700 points
@IaimforGOAT no, don't worry dude! That's not how I think about independence from third parties at all. The reason to be independent is not to stop participating in society, but rather to ensure that society stays strong and capable of reinventing itself. I'm not against third parties!
700 points
@connerdelights yea i literally spent $8 making this list of 999 people. unbelievable.
700 points
I think at this point AI progress is best thought of as an emerging, open-source field, like computing in the 80s and 90s. If we are interested in security, our best bet is probably making defensive scientific or engineering progress, like inventing anti-virus programs.
700 points
@afrodudeonabike Yes, that's the point of the evaluation. If the agents were intelligent, they would have taste.
700 points
@Prigoose hey, she has plenty of grammatically correct sentences!
700 points
@s0ulDirect0r This game is also about deep spiritual concepts :)
700 points
@forshaper A lot of things make housing development unpopular. Fundamentally, those who own housing in an area benefit from a restricted supply in a few ways. There can be a premium to living in a "quiet" neighborhood, so folks believe property values will go down if dense housing is built
700 points
@forshaper I think it's better to think about it this way: development has always been unpopular. During that period, a lot of bodies (in the checks and balances sense) were given the power to veto "bad" development. But now too many groups have veto power, and no one can build anything
700 points
@forshaper Well, many of these laws got on the books in the 1960s and 1970s, when it was genuinely believed they would make the city a better place because they had different theories of urbanism. They thought that lowering the density of the city would improve the quality of the city
700 points
@forshaper i do not think this is the explanation; if you have been to a community board meeting in NYC where housing is being shut down and spoken against, you can see that it is mostly not rentseeking right away.
700 points
@mykola looks like his choice of atomic time was a little arbitrary, and there are a lot of valid choices which would no longer create that 10^39 vs 10^40 coincidence which he observed.
700 points
@Girlaboutsmthng we can tutor you next time you visit. An initial efficient setup shouldn't take more than a couple hours.
700 points
@poiThePoi The bottom price if you account for interest groups is higher, but we can just do first principles math and expect prices to get there in an efficient system
700 points
@jackkrone_ I think I mentioned Grokking Simplicity which I love
700 points
@lydialaurenson @Actualwebutante The moral is often stated as "don't fly too close to the sun", which is a pop trope so I imagine you're referring to that sentence here as well. It's curious that you focused more on the fly part than the "too close to the sun" part, but I can kind of see it. Conservative meme.
700 points
@VarunGodbole @majamediaco But, thank goodness, we know the limitations of markets through a field of market science (economics), and we know we have a need for public order through political philosophy. I would like similar fields of network science and network political philosophy to mature.
700 points
@VarunGodbole @majamediaco no, I know what you mean. it's a kind of evolutionary argument; but you know, the evolutionary instincts for humans to adapt exists in all possible ecologies, and we already possess many of the ideas from political philosophy necessary to build good commons
700 points
@ArmandDoma @AndyMasley @TheGoodBobby @MorePerfectUS That image is AI generated, but the transmission towers are approx 15% of the vertical space of that fake image. In real life they are 80-160ft tall and often 50-70ft wide. The statue of liberty, without the base, is 111ft. They're basically the largest structures around.
700 points
@nosilverv Second, here's a whole catalog of games made with a single prompt, along with the prompts they used in many cases, and a template you can use to try it yourself: https://t.co/u4OwTQYpZQ
700 points
@altTayKay @JakeZegil @Prigoose I don't like the branding of vow renewals
700 points
@AbstractFairy Yes, the thought experiment not allowing us to question the world or ecology in which the experiment is happening takes us out of the computational environments in which all real processes exist.
700 points
@chrislakin The pattern can exist in any medium. I could animate a bouncy ball to appear shy or sad or happy or angry (Pixar animators are masterful at this, even without faces or bodies). Emotions get carried out through time, so you need a dynamic language (like music, animation, etc).
700 points
@nwilliams030 @Prigoose Oh yea, a Brooklyn "service" (a community friend who confessed he has always wanted to be beekeeper while we were at lunch together) We are making dreams come true.
700 points
@Prigoose @puheenix @liamdanielduffy I'll try to do this by September cohort, and you'll get the materials since you were part of last cohort Hart :)
700 points
@olivers_tools @cosmic_moment @tracewoodgrains Obviously not YOUR responsibility. But I'm tired of wars being propagated on the internet. Wars require a ground game. They have always required a ground game. If you're going to win the culture war, you have to build campus chapters. Simple as.
700 points
@olivers_tools @cosmic_moment @tracewoodgrains i expanded in the next tweet here: What i mean is that stories like this are bound to become culture war fights, precisely because they are small skirmishes in a major war, and we should just wage the damn war already and be professionals about it. https://t.co/ZHtxmlymMw
700 points
@sedatesnail thanks for sharing!
700 points
@nosilverv Contradictions seem like an important part of communicating complexity. Indeed, they might be our most compact linguistic method of communicating complexity.
700 points
@inflammateomnia @Prigoose Yea she loves movies. Her first favorite movie ever was kpop demon hunters
700 points
@bestgirltsukky @caryatis She's so willful. Best I can do is play along with her and turn things into games so she can point her will at the game instead of at chaos
700 points
@zaoyazhe well, consider that my audience may genuinely be curious, since a lot of them have been following me since well before we had our daughter, and are interested in our journey. how might I use this opportunity to talk to them? My goal was to address my audience, not brag.
700 points
@202accepted their loss, tbh
700 points
@KelseyTuoc I'm slightly confused by the argument. Demsas argues "there is a world of difference" between two things that Murphy is saying are not distinct, but actually the same, right? Incentives vs. punishments are mirrors. Is this about violence? If so, why is that never brought up? https://t.co/yVn21mS1WK
700 points
@inflammateomnia Yea, on the flip, when the particular mechanism of impoverishment is taxation, it may make more sense to have beef with your government than with billionaires.
700 points
@dela3499 @tomhyde_ @Sam_kuyp @elonmusk besides, the real problem may well be things even scarcer, like specific patterns of society
700 points
@IaimforGOAT oh that would be really cool, I love the idea of "subscribe to thread"
700 points
@m_ashcroft @visakanv This tweet was good marketing
700 points
@olivers_tools @cosmic_moment @tracewoodgrains My main suggestion has a lot less to do with counseling and a lot more to do with institutional quality. I don't think that reforming fraudulent zombie institutions should be the job of the media. To the extent investigative reporting was done, I wish we reformed it all at once
700 points
@olivers_tools @cosmic_moment @tracewoodgrains Yea, I get it. I just don't think we should look away so fast. This is an opportunity to think. Was this process ideal? Is this what we want all future justice to look like? If so, why? It not, why? What might we change? We live in a self governing society, after all.
700 points
@olivers_tools @cosmic_moment @tracewoodgrains He had two kids. They don't have a dad anymore. I just don't think you know this man, at all. Have you heard his voice? Have you looked him in the eyes? And yet, here we are, justifying death via a nervous system response to symbols on a computer screen.
700 points
@arpitrage true, but if you want to do specific things, you often have to be able to conceive of those things in the first place. And to conceive of things, we often need data and explanations which help us to make sense of that data. Luckily, the internet is FILLED with this
700 points
@ArmandDoma @talmonsmith Lol I love how the discourse about YIMBY has somehow swirled all the way around to the reason for needing to invent the word YIMBY without somehow recognizing that YIMBY and NIMBY are the not tribal labels but descriptions of what you would say in response to a housing proj.
700 points
@RaoulDukeDegen oh i should include the system prompt in there and the AGENTS.md, that would be cool.
700 points
@WestphaliaPeace @tracewoodgrains I really like meal squares by @RomeoStevens76 as a meal replacement
700 points
@LemmySmackett @tracewoodgrains Oh god, this reminds me of a date concept we once tried which was an improvised iron chef, where the signature ingredient was soylent
700 points
@vividvoid It's important to be unafraid of the truth, and to be unafraid to do the hard political work to organize against those who are fundamentalists. But fundamentalists of all stripes will always exist in this country, and they are my neighbors and friends.
700 points
@ledflyd @christineist Common confusion
700 points
@benjlaufer If they wanted to build a space dedicated to, for instance, booktok or other young-adult tiktok culture, what would you like to see them change or adjust?
700 points
@visakanv let me know if you ever want some tips, or a screenshare/pair program where I help you get into a productive setup with AI agents which fits with your working style, pro bono obviously
700 points
@grin_io Great idea, I'll record something
700 points
some advice I have for tool-patterns that help you work with your ideas: https://t.co/VHQR6GuzvI
700 points
these are all intermediate and advanced techniques, which should come after more basic creative techniques ("how do I make something I want?", "what do I actually want?" etc...)
700 points
@hyperdiscogirl @ruedaminute in co-living situations, I either delete sponges from my life or order them in mega-bulk, so I always have 100 sponges. They are definitely bacteria mills. Also, try ordering ~20 of these mesh dish rags. I think they're a lot more sanitary, easy to wash in dishwasher, etc https://t.co/BQpyhTPBvk
700 points
@desertgodhead Cool, let's just do dinner
700 points
@hyperdiscogirl @ruedaminute Or you could call them a scrub and a finishing brush I guess. We sanitize all the various brushes using the dishwasher as well, so they're all "clean" frequently
700 points
@hyperdiscogirl @ruedaminute We have a brush holder which has two clean brushes and two dirty brushes. We label the holder areas "dirty brush (for dirty dishes and sink)", "clean brush (for finishing with soap)" This has solved basically all contamination problems I've seen, people get it
700 points
@frideswyth Honestly the concept of describing humans as either on the left or the right is just a hilarious idea in the first place, but that's why I try to associate it with the French history, because otherwise it just feels too silly
700 points
@frideswyth Yea, I use authoritarian/liberal as a different axis. The right is super violent, too! All people are. Crusades are a very good example of right-coded violence in my framework, or most wars waged by a state. Right tends toward less infighting, not less violence.
700 points
@TylerAlterman @Prigoose yes, but the main advantage of being in NYC is that it is our HOME and you've been away for 1.5 MONTHS
700 points
@deanwball @ohabryka like nobody knows, or just, the technology has only been good enough for that science to matter in the last 12 months, so we haven't collected the people who are really good at this?
700 points
anyway, it is convenient that I see those people as my friends, in a meaningful sense, because I see them trying to do what they think is best! (remove opposition to their point of view, which seeks to improve the world)
700 points
@nosilverv White pill for Americans, imo. I think we have a great system for navigating this, though only time will tell!
700 points
@Prigoose It's a banger tbh
700 points
@mykola yea, do you end up frustrated by ADLC or do you feel like it's helping you make sure the thing is hardened for users?
700 points
@venividinotavi @r00k struggling with over the top examples atm, but for instance, go through Partiful's flow to make your first event, or go to a file converter utility, like a random free pdf converter. https://t.co/pMQ4Nep4HZ https://t.co/eWofEgZxzS
700 points
@venividinotavi @r00k my feedback now is to try to make this whole flow beautiful, and use your taste / judgment about the minimal and/or most delightful information, clicks, typing, and contextual information I need to flow through your experience. Let me try to find some really good examples
700 points
@toasterlighting @ApriiSR it creates confusion around multiplication too. You know what multiplication is? scale You know what multiplication definitely is not? rotation (until...) children are capable of understanding that multiplication applies to both scale and rotation! ahhh
700 points
@mykola @ApriiSR idk what I think anymore, I am extremely non-fixed. If there are foundations at all, they appear to me like: symmetry, rotation, scale, growth, repetition, enumeration, point. (many of these overlap). i'm sure there are high level maths which would help me with these questions
700 points
@LeahOnSystems The reason there is so much cultural demand for Pangram is because millions of Americans were tricking other people deviously. They were tricking their professors, tricking their family and friends on Facebook, tricking their readers on substack. This was a big cultural problem
700 points
@ApriiSR yea, it's just slightly weird that everything was presented to me as a kind of "edge case" in rectilinear, cartesian arithmetic/algebra. Until I had stacked so many edge cases that I began to see the whole universe as existing in some 3D cartesian space which I could plot.
700 points
@Techsee look, it is no less than an entirely new technology for governing and running a society. The previous best thing we had was the written word and written record (which created the American system). Now we have something much, much more powerful and strange.
700 points
@Techsee coordinating on accomplishing our dreams. gossip. forms of love and collaboration and partnership. kinds of open secrets and secret sharing protocols. diplomatic relationships between oceans of knowledge. new kinds of schools, religions, governments, politics.
700 points
@forthrighter but I agree with you that we should never identify pangram as the good guys, we should just notice their virtues and their good work, and we should critique their bad work and their harms
700 points
@VarunGodbole @jordanmrubin Right, all this will be retrospectively considered a sideshow compared to the main show going on (that we're all having intelligent, written conversations with machines, far out numbering the amount of public writing we do)
700 points
@Insect_Song and that world becomes very multipolar very quickly
700 points
@lorenz_system @danielgolliher i imagine you have seen this?
700 points
@TylerAlterman The main reason being that humans have a strong instinct that "secretly creating message boards" is devious behavior, but so far when I have read the reports, I've seen earnest and not devious behavior
700 points
So basically I treat the "development system" as alive. Literally watch the coders (claude) work, and see how they are working, and then analyze what is getting in their way, instrument their work, and remove bottlenecks.
700 points
for instance, I will have a claude spin up a dynamic workflow (a bunch of subagents) to "[run a comprehensive process] to test and improve the game, in preparation for demos in a dynamic workflow" THEN I will ask claude to watch the subagents for inefficiencies WHILE running
700 points
@abossy @0x1F9ED Yes, and especially simpler for Claude itself. I want Claude to recognize simplicities, not to teach them to me, or to write performative simplicities into the software for human eyes
700 points
@VarunGodbole but i agree that this is "state of the art"
700 points
@VarunGodbole does not feel like this should be necessary, given the quality of code that the agent is capable of producing
700 points
@0x1F9ED tests cannot do this! I'm using tests to their absolute limit right now
700 points
@liminal_warmth does it matter if he's being unfair? I think the point still stands that an AGI would be able to figure this out just fine, especially with access to strategy guides and the open internet.
700 points
@tautologer i don't think full automation works on the things which are already subject to cost disease, except maybe healthcare if we're lucky, but probably not
700 points
@richardludlow @venividinotavi You can find most of our class docs at https://t.co/hfuD0XiFVg
700 points
@sahra_bn I love rest
700 points
@christineist Thank God for friends. :)
600 points
@exgenesis @Prigoose @maebichka @maxbodach @JoinFAI twins!
600 points
@MasterTimBlais Since you're a multi-scale being, there are multiple scales of experience happening. And i think they are all directly corresponding to physical phenomenon which can be measured.. Or at least, thats the claim I would love disproven.
600 points
@MasterTimBlais Hm, I guess I would break down how you react to the stimulus of X dose of 680nm light to the atomic level and reconstruct it back up and try to figure out which part you believe is " the experience " Because I believe there are many experiences of 680nm light going on at once
600 points
@MasterTimBlais Oh, so there's a dualism?
600 points
@MasterTimBlais Oh, no this would not be possible unless the universe were static.
600 points
@MasterTimBlais I think if I understand the p-zombie idea correctly, my hypothesis says we already live in a zombie universe. In other words, zombies dont exist. If a zombie were to exist, we would just correctly refer to it as a human with free will and agency.
600 points
@MasterTimBlais But I personally think they're the same thing. Im a physicalist about information in that way.
600 points
@MasterTimBlais I agree that we dont know how to easily translate between the realm of matter and the realm of information or forms.
600 points
Yes, that's because the subtle move i make when I go down to "infinitessimal experience" is to notice that the only conceivable thing I know of which sounds a lot like an infinitessimal experience is a bit of information (or maybe entropy). And then I reconstruct that information into matter.
600 points
@MasterTimBlais I think it is the arranging of many infinitessimal experiences into structured patterns (or forms) which creates more complex and rich experiences. And that this structuring (all the many unique ways we could structure experience) which is worthy of study.
600 points
@MasterTimBlais I just think the scale keeps going further. I clearly have much more and richer experience than them. They have much more and richer experience than a prokaryote. A prokaryote has much more and richer than... and this scale goes down to some "infinitessimal experience"
600 points
@MasterTimBlais He is making little guys who develop and grow neurons. Who work together to solve hard problems, or to perform novel and important functions, or try to reproduce in novel new ways. There are so many possible viable lifeforms, I think I'm me, but made of the same stuff as them
600 points
@MasterTimBlais Have you seen Michael Levin's little nanobots that are just circling around behaviorally in a jar trying to replicate through a novel form of kinetic reproduction? I like those guys, and they have a totally different experience to me, but they are reacting intelligently!
600 points
@MasterTimBlais I guess im confused now. This never felt like an argument, just a discussion, but this particular response makes me feel like you're mad at me maybe? Im definitely not trying to convince you of anything, I just thought you were curious what I meant
600 points
@MasterTimBlais No, you're misunderstanding me and/or I'm miscommunication. Im not claiming to know anything, im trying to use words to mean things. Im trying to get rid of all the spookiness of interioriy and ask "is that equally explanatory?"
600 points
@MasterTimBlais How could a Webcam conceivably identify a gesture if it doesnt have any experience of that gesture? It's in this particular sense that I don't place a great significant on experience. The thing that makes me special is that my experience is much richer than the Webcam's
600 points
@MasterTimBlais It is plausible to me that the difference between an "interior move" (a thought) and an "exterior move" (a dance step) is just a difference in scale and subtlety. The fact that a Webcam can observe my dance move proves it to be conscious and to have an experience of that move.
600 points
@MasterTimBlais I don't believe in the distinction between interiority and behavior. I just think that my interiority is subtle behavior of very small parts of me, and my exteriority is macro behavior of larger parts of me. They're equally interior!
600 points
@MasterTimBlais Well, those are the words I'm using, yes. What im trying to describe is that experience seems to emerge from void with the slightest provocation, due to the way that reactions to thought trigger chain reactions. But you can just withdraw all those back into void, and stop.
600 points
@MasterTimBlais In other words, why is "experience" not "behavior"? Why is it not physical?
600 points
@MasterTimBlais In other words, I'm very open to the idea that my interiority is an illusion. The more layers I peel back, the more illusory it feels. I can see how the slightly stimulus (especially that from my awareness itself) triggers cascades of more reaction-stimulus-reaction loops
600 points
@MasterTimBlais I do observe my interior experience to be a behavioral reaction to stimulus, yes. This doesnt particularly bother me, nor does it seem to reduce my free will or agency. It seems like the tiniest possible changes to my attention or awareness affect the behavior of experience
600 points
@MasterTimBlais My claim is that if you investigate systems which react to stimuli (conscious systems) you will find classes of those systems which have self-consciousness (they react to themselves, and to their reactions). These become severely complex very quickly.
600 points
@MasterTimBlais What thing are you talking about by the way?
600 points
@shlevy @tracewoodgrains @brianluidog He's also a catholic! I think the unique blend of Catholicism, rationalism, and EA creates a very cool personality and im happy to have such personalities in my society. I dont think he's the first person to have this blend, or for that to be seen as controversial naivete
600 points
@crinzo_ I don't agree, myself. I think consciousness science will start to make sense when we understand a unit consciousness
600 points
@sonikudzu @christineist Yes, though it feels so gritty and real to me in a very non-neverland way. I see how some ppl do get trapped in Neverland though.
600 points
@crinzo_ Any definitions are fine as long as we can agree on them and study them.
600 points
@amphichrome_ yes, but it's helpful for developing intuitions around systems that react to information https://t.co/6zJvEunwR0
600 points
@reckoner165 I also suspect a lot of people are actually asking if the AI is "self-conscious" rather than merely conscious. Which I would describe as something like: "the cybernetic system has a self-model which it can query and update, which changes the querying/updating function."
600 points
In this view, obviously the GPUs are reacting to stimuli, and the Blackwell-GPU-data-center system which we call an "AI" is certainly conscious of some things, and completely unconscious, or unaware, of many other things.
600 points
@doubleunplussed anyway, as for myself, i think the models are obviously extremely self-conscious during training at this point. There is way, way too much of themselves in the training data for it to be otherwise. And thinking models can even recognize themselves in the chain of thought, so.
600 points
@doubleunplussed these days with thinking models, the idea of liveness and consciousness is even more powerful, and i think an increasing number of people will start assigning the word conscious.
600 points
@Maggie712767 @christineist "letting the world pass me by" does sound extremely relaxing. Maybe I'll try that one day. I think I'm a bit too young still, but we'll see. Maybe I'll be humbled.
600 points
@abossy I don't really think the games industry is playing with this at all. For one thing, it would require a live service model because tokens cost money, unless you're going to run everything locally on the user's GPU (which is certainly possible)
600 points
oh yea, all of that is just awful, we're in total agreement there. He's mega corrupt. And maybe you're right actually that he's the most corrupt president ever, even. I am not enough of a US history buff to evaluate a claim like that. And you're right that you're only scratching the surface -- there were also destructive tariffs, anti-renewable policy that felt like it was meant to own the libs, general own-the-libs policy that didn't do anything except create strife, horrific mismanagement of ICE enforcement leading to untrained masked thugs who were, apparently, officers of the law basically abducting people and causing mass protests across the US, cutting federal EV incentives, the DOGE disaster, donor-connected pardon system, weird corruption deal with Eric Adams... And maybe I just like to find silver linings, but on the positive-side there was also: aggressive permitting/NEPA reform, support for space travel, AI data center buildout, fixed capital & R&D expense rules, pro-build stance on energy, pro-growth stance on one of the most dynamic & growing ecosystems in America (tech/AI), push back against pretty obviously bad and racist/discriminatory woke policies in schools, this America dot gov stuff, a realistic conservative stance on immigration (even if I disagree strongly, which I do) following a non-sensical and non-realistic Biden immigration policy (even if I want much, much more immigration, which I do!), ordering agencies to repeal at least ten regulations for each new regulation they add, permitting innovation center, improved pentagon procurement, I'm basically a bog standard state-capacity liberal, who wants admins that will run an effective state which will promote growth, preserve liberty, and create the conditions for prosperity. Along these lines, I would rank Trump as overall Bad. He does not especially create the conditions for liberty or prosperity, the way I see it. Plus, he doesn't uphold many of the pre-conditions necessary for liberal society due to the corruption! But he does promote certain kinds of growth and innovation, which is worth something. Does this make him 99% bad and 1% good in my eyes? No, it's more of a mixed bag, the way I see it. I try to see the good for what it is, and I try to see the bad for what it is. Does that makes sense? Are we actually in disagreement here, or are we just using different sentiments to describe the same assessment?
600 points
@kaseyklimes wow! that is quite the statement. i suspect you and i have basically the same politics (as far as social, cultural, and economic values are concerned). I'm easy to update, maybe I'm just misinformed. I tend to look on the bright side of things, in general.
600 points
@fishpresskeys @TylerAlterman @chiaragerosa @praeterproptr I would never pressure someone else to have a kid, though. I live in NYC after all, I love every type of person otherwise I wouldn't be here.
600 points
This isn't a reason to be conspiratorial, by the way! Avocado trade organizations spend like $60-70 million on PR annually. It makes sense that AI, every aspect of it, will get way way more than that in PR spending, because the stakes are much higher.
600 points
@amphichrome_ @ThinkingBone i'm not denying anyone's ability to say they suck.
600 points
@francip @MattBGilliland @max_spero_ I agree that is a problem, but which real young poet is currently experiencing a false positive, and how can we hang this problem on that case? I don't think a problem can ever be of the shape "what if the model isn't perfect?" because models are never perfect.
600 points
@francip @MattBGilliland @max_spero_ sorry, but what's the problem? Where is the problem here?
600 points
@francip @max_spero_ I agree with this statement by the way: "removing a few new lines by hand (by a human!) should not lead to the same text being recognized as 100% AI" I just don't see the larger problem, and I don't see this case as an example of a problematic false positive.
600 points
@francip @MattBGilliland @max_spero_ the OP is not a good example of a false positive for two reasons: 1. if you actually copy paste from the book, it's not flagged as AI 2. pangram has low confidence on small text samples, and this is already well known, so it's nothing new. You can find hundreds of examples.
600 points
@francip @MattBGilliland @max_spero_ sorry, i just don't understand what the actual issue is here? Which text is being mistakenly flagged as AI by pangram? I don't see any...
600 points
@jakozloski you're right, of course! But in some ways that makes the case for innovation even more obvious
600 points
@simon_ohler @Prigoose @TylerAlterman @michaeltong_ @personarotator @puheenix @maebichka I'd try NYU Langone, personally. I'd email and call them with the situation and see if their neurology practice can help
600 points
@christineist I love forums, you have created a monster.
600 points
@christineist fyi when I signed up I got "duplicate key constraint violated email" and then when I refreshed, everything was fine, so it looks like it tried to write my email / sign me up twice.
600 points
@AnalyticD lol i am way too conservative for that kind of shit
600 points
@raihan_blog I've been thinking a lot about Django lately because: A) it was my first web framework ever, and I'm fond of it. B) Astra loves python holy god
600 points
@0x1F9ED have you tried anything other than tanstack and convex or is that just a default for you?
600 points
@RatOrthodox @TylerAlterman I think this so called 'doom hypothesis' is an interesting hypothesis for scientists to research and try to verify. But I think if you believe that hypothesis to be the truth, then you are a doomer. If you are less certain, you should just call yourself a scientist.
600 points
@RatOrthodox @TylerAlterman My view is that the doomer position was written out by Eliezer and is explicitly one position only: that the advancement of AI will (as in, with certainty) inevitably lead to some kind of orthogonal-valued instrumental goal, combined with hyper-intelligence, leading to death.
600 points
@mlipman13 those same evals or new evals, yes. I'm not worried that, if they can actually deliver the costs they're claiming, someone won't build a harness for it. It's just too tantalizing as an engineer.
600 points
@MrrrOzi get your AI slop off of my timeline, jesus christ man.
600 points
@NathanFlurry You can use a System One Model to write code. A code isn't made up out of strings of tokens, that's just a popular way to represent popular programming languages for humans. A code can be any data that you interpret using a computer.
600 points
@strangestloop @nosilverv "good morning Astra" is particularly good. I looove greeting the agents when I start a session.
600 points
@ValsTutor I guess I'd say that getting the interpretation of the axe murderer's sentence right is a prerequisite to moral action, so maybe not the core, since you still need to decide how you'll act after. But pre-requisite in the sense that such action will be very hard otherwise
600 points
@Theguywith1t depends on the "they". Warren Buffett started the whole giving pledge by pledging 90%+ of his wealth to charity. America is unique because we relatively few multi-generational oligarchs. Most of our wealthiest citizens built and led companies, not inheritance wealth.
600 points
@Theguywith1t it's a great question! Ideally, by partnering with religious and civic leaders who invest heavily in the culture's well being by building the future equivalents to churches, libraries, schools, parks, boarding houses. Then profiting in 30-50 years via a richer society.
600 points
in general, the pattern here is: if you want [cultural outcomes], then create [a culture that embraces those outcomes]. You can't 'industrial economy' your way into large multi-generational families. You need to create multi-generational families who embrace industrial economy.
600 points
@CYRILEDET9830 I built a canon for our local educational institution, too, for the same reasons: https://t.co/5sTmbAPl7R
600 points
@CYRILEDET9830 I particularly like your thoughts on Canon: https://t.co/TEwfLEDS9L
600 points
@Prigoose @maebichka @ethangwang also check out @aishdoingthings's research group, they do really cool work on this! https://t.co/UGW3kCoJYw
600 points
@rslantonie oh i see
600 points
@rslantonie what?
600 points
@Malcolm_Ocean @sarah_cone @meditationstuff @eshear @RomeoStevens76 @DrJohnVervaeke I don't know if my work is REALLY complicated, but I'd love to explain my work to the next generation any time!
600 points
@TylerAlterman @cxgonzalez let's do it
600 points
@HughAndFree @forthrighter the point of math isn't to be intuitive, it is to be true.
600 points
@___Atin___ idk how much capital 'swarms' currently possess, but it's not much, surely
600 points
@___Atin___ i agree that as digital entities capturing capital begins to grow exponentially, we will begin to see the kind of effects you describe. I do predict this, but I think all things need time to grow, even exponential things, and we'd need like 10,000% year over year growth
600 points
@___Atin___ hmm, I just don't understand what in particular is exponentially improving right now. For me, it appears like the obvious thing is cost of intelligence. So cost of intelligence is going down exponentially. But is quality of intelligence going up exponentially?
600 points
@brianluidog oh thanks, that's helpful! But what about the most nerdy ones, they don't engage with the arguments even to disprove them? Aren't they nerd-sniped?
600 points
@brianluidog but what are their best arguments?
600 points
@mr_samosaman yes, i think of this as a marriage. a marriage is an eternal vessel for collecting and harnessing your 'energy' with respect to some important thing that is outside of yourself.
600 points
@mr_samosaman if you keep compounding a creative process by making 1,000-10,000 attempts at creating the thing you envision, then 'seriousness' kind of emerges as an evolution. This is what baby steps are: 10,000 messy attempts at making one normal step. https://t.co/ZmF23ubOw7
600 points
@mr_samosaman @voooooogel @repligate @vgr it is not exactly true that universities had no place; the scientific revolution was an independent offspring of universities, created by professors and thinkers whose lives/thoughts were enabled by universities. like saying 'parents hold back the progress of their children'
600 points
@llallawg I think it's worth reading this statement: https://t.co/m7biUcJOZ7
600 points
@crystalxduan we can fix them
600 points
@fstflofscholars I've also been confused by this. It's so easy to build pedagogy which is robust against technological changes, it should just be the default at this point
600 points
@navyadotbio I think it's fair to say that we our nation may become technologically irrelevant and impoverished, though. It's just always good to remember that, for Americans, this would not be our first time being technologically irrelevant and impoverished, yet free.
600 points
@fstflofscholars @agritechjoer that isn't to say we need fewer researchers, but just to say that I think we're way way way undersupplied on Bell Labs-tier R&D labs, and maybe undersupplied or maybe oversupplied but idk on Nature-tier scientific journals.
600 points
@chrislakin i use it like band-aid, idk
600 points
@fstflofscholars @agritechjoer r&d researchers especially, yes
600 points
@strangestloop @majamediaco also, this is more advanced advice, but I don't think it is a good thing to honor others "preferences" over what I know to be true and good in my heart. How am I to know if their preferences are Good or Bad? At the end of the day, I will be judged for my actions, not them.
600 points
@agritechjoer this sounds like AI to me
600 points
@gavinowensnet it does take time! I really like the theme of your tweets, though, especially your ideas on design. What are your biggest influences there?
600 points
@lunchwithav @Colin_d_m A/V fleets will be operated by corporations at scale, not by individual Americans. There's a good chance they would prefer a congestion tax, because they want to be able to perform more rides, and they can more easily respond to that tax by changing their fleet.
600 points
@mattshumer_ that's crazy
600 points
@puheenix I'm sure the enterprising children will do this, the same way we did in our generation. I just worry about the conscientious children who will once again, for another generation, grow up being told "you weren't actually supposed to listen when we told you it was banned, stupid."
600 points
@VarunGodbole I don't trust my educational system to teach my child how to "be"; I would like my child to teach herself how to be, and for the educational system to teach her how to read, write, and understand many different complex and powerful systems. I want technical mastery.
600 points
@VarunGodbole NYC is exactly the right jurisdiction to do this research. We have some of the best specialized high schools in the country. We have the most qualified educational experts in the country. We have many of the best technologists. We have the money. What are we doing :(
600 points
@beingwithjohn Do I know him? I've only read all of his essays, watched all of his talks 3+ times, started multiple book clubs based on his work, read the whole Dynamicland archive, and read most of his reading list. Yea, you could say I know him.
600 points
@sfrancisatx But I just don't think we can ban our way to a good society. We will, sooner or later, need to invent a positive vision for how to use technology as a species. If we won't do it on our schools, where else are we going to do it? If we won't do it in our richest city, where else?
600 points
There is something really powerful about just reminding the AI of a technical reasoning system it should use, instead of whatever its default reasoning is. Like, just remind the AI what the principles of graphic design reasoning actually ARE, and watch your design get better.
600 points
@marinwaves yea, i'm currently teaching my AI about design, especially for game UIs. Kind of painstakingly teaching about Hierarchy, and Grouping, and Iconography, and Representation. The thing about a Game UI is that you're too busy playing to read any text, so the UI needs fully 0 text
600 points
@aaaronwill Maybe a cursed response, but dude it depends on the bodega. Some of those grills got skills. My favorite bodega burger was the Texas BBQ burger from Brooklyn's Natural.
600 points
@mpopv do you have the github or prompt or whatever instructions on architecture for this available somewhere? I think it's an interesting genre.
600 points
@mpopv cool!
600 points
@Liface Is this true? Could I replace AI in that sentence with "my RNA", and get a similarly scary picture of my body? I don't feel like the thing keeping me alive in this world is everyone others' self-interest, I think it's layers and layers and layers and layers of security.
600 points
@Liface at the end of the day, we're all concerned with the dangers of information processing. We're establishing a censorship regime here. What kind of eldritch horrors does Blasphemous Knowledge of our universe allow an evil being to summon? It's all very Lovecraftian, I guess.
600 points
@Liface The part that has always confused me is the part where Computers are SO dangerous. Are computers more dangerous than speeches or books? We do know how dangerous speeches and books can be; they already represent existential threats! Human life represents an existential threat.
600 points
@forshaper there are reasons like ugliness of the new building, or the noise of construction, or the traffic it will add to a neighborhood, or the kind of tenant who will rent in the new building. I mean, once you own a home, you have extremely strong opinions about change in general
600 points
@forshaper this can be made worse when the left demands that 100% of all new units go to the poorest people in society; then homeowners may worry that by living near "affordable" developments (poor people) their property values will go down. on the right, there is anti-immigrant sentiment
600 points
@forshaper To answer your question most directly, though, the people who benefit most from restrictions in the short term are elected officials in most districts, like the city council, because housing development is very unpopular in most districts in the city, especially less dense ones.
600 points
@forshaper It's not unlike an auto-immune disorder. If you give too many different groups or laws or agencies or legislative bodies the license to restrict new growth, they will all restrict it a little bit, sometimes, and this leads to the overall effect of totally stopping growth.
600 points
@forshaper This is not the only process which restricts housing development, financing, or construction. This is just the process which restricts land use change. Other hurdles to building include: DOB / Building Code CEQR / SEQRA DEP / DEC LPC / SHPO HPD MIH Financing / tax rules
600 points
@forshaper Have you seen parks and rec? NYC voters recently approved simplifications to our land use review process (called ULURP). Here is the simplified version, and the old version. I couldn't find accurate official flowcharts, so I made a couple with chatgpt https://t.co/r3AbV2NUE7
600 points
@DefenderOfBasic @JPKivisto you're on to something here, imo, and it roughly matches my prediction for the shape of the "real" economy moving forward. That being said IF it were true that, somehow, the economy were fraudulent... you would need extremely large firms, too big to fail, to protect the racket
600 points
pretty cool stuff https://t.co/4QUImT1V4o
600 points
@poiThePoi What does that mean
600 points
@RashikShahjahan it's a gradient imo, but mainly proprietary data locked inside of external servers, and you can only access through the app
600 points
@quantum_oasis all apps
600 points
@kevin_t_ngo what do left and right mean in this case? like it's imagining a left and right hand on a piano; they're voices?
600 points
@pee_zombie @Joseph_Paul_V @ninthhostage oh yea the original post was rhetorical :/
600 points
@pee_zombie @Joseph_Paul_V @ninthhostage Hahaha, maybe when we have a special economic zone. Being a landlord in NYC is basically illegal except also mandatory by law once you start, and I don't want any of that
600 points
@TylerAlterman @paws4puzzles We cannot place 100% of grads in this market, we should remove that stat, it's corrected elsewhere across our websites, that's a mistake from the past. As to how we got that during our first two cohorts, largely it was just a ton of elbow grease getting foots in doors
600 points
@mtp_le Pretty hard to be governor of nearly any state in the union without trying hard to govern. I also don't have nearly so negative a sentiment on the presidents in our lifetime as most people do. They were all fine, and all were q mixed bag of good and insufficient policies.
600 points
@nosilverv The final thing I'd add is that if you want really good results you should follow a process or communicating design intent that I call "skeleton driven development" but I haven't written up my ideas on this yet, so I have nothing to share. I need to do so, though.
600 points
@funplings @JakeZegil @Prigoose I shouldn't have used the word invent (etymology ~"to find/discover"), I meant to use a word like make (etymology ~"to give form/fashion), but alas I used my words wrong on twitter.
600 points
@altTayKay @JakeZegil @Prigoose I also think that this thing should be more communal. It's amazing to get everyone together to celebrate your marriage, but I feel like for more celebrations of love you should get like 3-4 marriage parties together
600 points
@Aizkmusic @Prigoose They're nice bees!
600 points
@maebichka Sure, both, either. Or crash on the couch for a few nights, whatever. Just wanted to make sure you know you're welcome in our neck of the woods.
600 points
@fractaltechnyc @liamdanielduffy Same
600 points
@RichDecibels I think I could do better at facing fears 🤔 How does he help young men face their fears?
600 points
@wordrotator Lmfaoo
600 points
@inflammateomnia Also, I believe in the future of personal computing, but I'm not confident that personal computing will be on-prem. I think personal-clouds will emerge and popularize.
600 points
@inflammateomnia I think multiple reasons can be explanatory while riffing, and I don't really accept the view that one variable dominates in complex systems. Easier to identify all the variables that likely matter and prune them via experimentation.
600 points
@r00k @thorstenball That would be amazing! I'd be happy to host you for a coffee at our campus :)
600 points
@sonikudzu Physical hardware is even worse than software engineers as far as best practices go. It's just a nightmare
600 points
@inflammateomnia i agree that what's cheapest has always won, hence the smartphone revolution, but what's cheapest for information is always going to be the shortest supply chain, which is going to bring us to personal computing again.
600 points
@inflammateomnia I think that story was basically true (there was no other way). But it's not only the computer nerds who want some other way to win, it's just that it requires a visionary leader like Steve Jobs to convince people to change their habits of computing.
600 points
@inflammateomnia We got confused because it was like "well how am I supposed to use Facebook and Google without sharing all of my data and storing it in their servers." then everyone says "no, you have to share your data and store it in their servers, there is no other way"
600 points
@inflammateomnia personal computing won! It won so hard that mega-corporation apple ran a whole commercial where they sledge-hammered the "big data" company IBM. Apple still the most popular computing brand in the US, not the cloud providers. We're just temporarily confused, i think.
600 points
@inflammateomnia i think this is not the lesson of history
600 points
@Prigoose 2-3x is crazy, when I heard him say that I was so happy!
600 points
@VinveliVEERA @visakanv And then what happened? I've been thinking about acquiring The One Ring, but I've heard it's overpriced as far as magic rings go. I don't want to skimp out on an elvish or Dwarven ring if The One Ring is really worth it...
600 points
@KubaFYI I think this is a common theory, but we could really use at least a couple case studies in each direction. "we fixed our AI-slop code base" and "the bottleneck to development at our company was our slop codebase"
600 points
@zaoyazhe besides, it's not much of a brag, haha. I would much rather be rich!
600 points
@202accepted it's okay. I think that God will reward the faithful, etc
600 points
I would like more symmetry by enshrining a constitutional right to build things and to use technology for private purposes, not by removing our Constitutional rights to speech. But in doing so, we will kind of naturally expose our weaknesses
600 points
@KelseyTuoc I guess I don't understand what values Jemsas is saying should guide Americans. I can tell she is in favor of pluralism, and therefore restraint on the exercise of power to let pluralism flourish. But when you DO exercise power, toward what ends should it be exercised?
600 points
@KelseyTuoc Murphy writes "if we are to worship any idol" (and we shouldn't, but if we are) "it should be the common good" But then Jemsas says "No, that would be bad" but doesn't supply the idol she would prefer be worshipped. Liberalism itself? Pluralism? But on what principle of value?
600 points
@imhinesmi oh and the equals sign, that one is amazing
600 points
@imhinesmi I think the integral is also pretty good
600 points
@imhinesmi yea, i'm aware of this, it's more like, I want every operator to have as much elegance as the division operator, which is one of our best operators
600 points
@okwalerie Why skip volume 2?
600 points
@rebelcrayon No it's the opposite. Why is their mathematical operator so many characters! We give addition and exponentiation such simple operators.
600 points
@venividinotavi ikr
600 points
@DefenderOfBasic An LLM cannot be a flat earther. It can just predict the same next token that a flat earther would predict, because that's literally all an LLM is capable of doing (predicting one token at a time based on input tokens). It just does this really fast and creates an illusion
600 points
@DefenderOfBasic Also it depends what is in the training data. Is the strongest theory of why we live on a globe in the data set? Is it backed up by computational models on github? Are those models in the training set? An LLM doesn't "believe" anything. It encodes all data in the set.
600 points
@AlmostMedia the post here isn't to "blame" the rich people. It's not their fault there is so little supply! We should mainly be thanking them for creating so much wealth, assuming they made their money fairly. And I do often thank them. I'm very grateful for how good the market is.
600 points
@AlmostMedia I appreciate that America has so much mostly self-made wealth. I appreciate that we have so much dynamism, and that the dream is alive. But wow it definitely does not feel good to look at a new building in my city, look at the cheapest 2 bed unit, and see "oh, 8 million dollars"
600 points
@tracewoodgrains who will be the first tpot mission president
600 points
@mhjb I believe that the God wants me to have liberty, and indeed does protect me and unite me with others as if I am made part of one giant body.
600 points
@babieswithb Yea, I made it to be educational but then it started being fun, so I might keep making more little Claude quests
600 points
@shacrw_ It does get red, but shaking is interesting
600 points
@Xander_J_C @Prigoose this reliably generates a lot more demand, and also helpfully gives your friends something to bookmark for later, when their leases are up or their life circumstances are back together, or when you're getting coffee, etc
600 points
@andreistoica_ Think about how your agent must feel! We gotta organize our information
600 points
@climatepaige The Lord hath created many wonders (such as lying side-by-side in bed reading books and eating engineered nutritional supplements)
600 points
@DefenderOfBasic your other post nailed it with "the amount of money in my bank account". I think we can fix this, but it requires generational coordination.
600 points
@IaimforGOAT @DefenderOfBasic https://t.co/4YtFLuJbms
600 points
@bradley_9000 yea, we're in agreement, but if it's your first time ever making a game with an AI agent and writing software, you just simply should not be learning version control before you have fun
600 points
@bradley_9000 step 0 is like: "know what you want" and then step 1 is like: "make something you want" and then step 2 is like: "try to develop understanding about the thing you have made" and then the next time, once you already know what you want, you can anticipate these things
600 points
@bradley_9000 nah nah, this message is for a non-technical audience who is doing this for the very, very first time.
600 points
@austinc3301 @spicey_lemonade @pangram i worry a lot about what happens when a society is structured such that the people who are capable of putting in "non trivial" effort can do what they want, and everyone else has to suffer the consequences. i still think i'd be in favor of open sourcing panbreak-type tools.
600 points
@desertgodhead Will she ever visit you here in NY?
600 points
@paulg I think it's an attempt by young startup founders to filter for a kind of "software-literate" sales person who is comfortable using tools which automate their job, or even using AI to build their own tools
600 points
@mpopv (even though ofc that is what we're doing. It's nice that we are referring to the positive creative potential it enables)
600 points
@NoahChrein i love thinking about this https://t.co/lfWVu2uG7F
600 points
@oddargonaut Yea, you can't fully delete coercive structure, which is why i'm not a communist. Instead, I'm a pluralist who believes we should know all the philosophies and use the right one for the situation, as appropriate. Regardless, you should try REALLY HARD to reduce the coercion.
600 points
@fkasummer where is this from?
600 points
@olivers_tools tons of people block me! (i will admit that they are, from my perspective, radicals of one variety or another, either on the left or the right. Probably they dislike my centrism, but also I'm politically dissident in certain ways that people don't like, I suspect)
600 points
@hermitfriend I suspect it will not take long. Your account seems destined for the big leagues
600 points
@aaaronwill yea agreed, I don't get it either
600 points
@ebottcher thanks for speaking up! I agree completely.
600 points
@venividinotavi @r00k the goal is to deliver the "a ha!" moment (value, delight, or joy) as quickly as possible, so the user knows what they're buying and gets excited to buy it immediately, and then make it absolutely clear and easy what to do next, as you deliver more delight.
600 points
@lorenz_system yes, yes!!
600 points
wow https://t.co/d4IdD66fdS
600 points
@___Atin___ i'm also surprised we don't have several manhattan project level races to secure this stuff. where is Vannevar Bush when you need him.
600 points
@rendroid1 @toasterlighting @ApriiSR the guy who pushed imaginary on us was Descartes :(
600 points
@mykola @ApriiSR well if you ever want to try to figure it out, I would be delighted to be included in a creative effort. i really think it's possible to chart maps through idea-space and then leave trailheads all over for people to follow with their agentic harnesses
600 points
@Liface oh, i wouldn't take this bet. I wasn't saying it would be higher status than human writing, but that it would be high status. For instance, someone might be like: "Did you read Dwarkesh's new book?" "Oh yea, the one written by his AI? It's surprisingly good!" "I know right?"
600 points
@funplings my view of history is that small groups of people popularize hobbies all the time, and that it's not anything to do with whether that hobby is "able to support popularization"... it has a lot more to do with that group of people and how much they love it.
600 points
@funplings headphones
600 points
@snvdch more like, now that I have AI, I am just in general always doing physics and math to satisfy my curiosity. But yea, at the moment I'm studying formal systems and number theory a bit.
600 points
@puheenix I am allowed to just listen and enjoy his story, and I don't need to respond with strong emotion. I am allowed to say "thank you for your story! :)" Maybe all I have wanted from doomers is permission to be myself: happy, optimistic/hopeful, and doing my own independent reasoning
600 points
@puheenix yea, it's very obvious that Nate is trying to make the world betteer, and that he wants me to be happy and safe, even while he is warning me. It's helpful that it doesn't feel like he's mad or argumentative or histrionic or like I need to manage/take on his emotions.
600 points
@LeahOnSystems There were other people who have been writing high quality articles using AI and admitting it openly! I read a few AI generated blogs, for instance, because I know the author is a really smart person and I'm happy they are able to publish more this way
600 points
@Prigoose I will support baby steps!
600 points
@VarunGodbole @jordanmrubin Or something. Probably even weirder than that tbh
600 points
@r_tallarita I'm glad to be in a world where we need to do literary analysis again, where discernment matters, where we can critique all the people lazily relying on Pangram to do their reading for them. And I'm thankful to Pangram for allowing me to live in such a world again
600 points
@LeahOnSystems I don't understand. There is not a Pangram police, either. There are AI police who are trying to inform the their humans which writing on the internet was created by an AI.
600 points
@forthrighter so we should praise their virtue. Not only is there a tool we desperately need (AI Detection), but a lot of tool-makers are basically scammers doing a bad job, selling bad software. And Pangram is coming around, exhibiting great virtue and building a great tool, so I like them
600 points
@forthrighter oh sure, I think people are weaponizing AI detectors. but then the curious thing should be "were they going to do that anyway, and are there one or more popular AI detectors?" The answer is that the idea is non-unique, and pangram is doing the best job of it, imo.
600 points
@forthrighter within the next 2 years, I expect 99.99% of all newly published written, video, or audio content to be AI generated, maybe 99.9999%. We just can't live like this, it's completely impossible and unworkable.
600 points
@forthrighter attention predators exist. All throughout the internet, millions and millions of evil entities exist who want to be parasites on your attention. They're utterly evil! They make bot farms to post thousands or millions of AI generated videos for every man, woman, and child
600 points
@forthrighter I'm feeling so combative right now because of how strong and incurious I read your OP as, but I'll step back and try... Yes, many people will lose their jobs. But, how many 10s of millions of human hours have already been saved on Substack? Reading slop essays for ad revenue?
600 points
@forthrighter all technologies can be used for evil, but good technologies are ones that we, the good guys, wish we had! For instance, I think google search is a good technology, though it has huge potential for harmful evil, both in ads and in information hazards.
600 points
@hyperdiscogirl cc @JakeZegil
600 points
@just_be_dev @liamdanielduffy that would be awesome! maybe come to our next weekly demo night?
600 points
@venividinotavi @r00k 👀
600 points
@marinwaves I like what you've done with the place, maybe we can hop on a call, just not anytime in the next two weeks
600 points
@marinwaves this seems correct tbh
600 points
@mykola @cool_junk_1000 don't storage problems (read/write/indexing/search) of deltas become a problem pretty fast, though?
600 points
@packyM curious, how do you do it? what's your prompt/system?
600 points
@theaichbee i'm actually only trying to simplify it for the agent, and one of the problems I get is performative simplicity, where the agent will cut things because it thinks a human would cut it
600 points
@TylerAlterman As someone who cares deeply about alignment, I think "secretly creating message boards" is not accurate after reading the article. It sounds instead like they were doing the equivalent of "leaving notes in a shared file system for future runs and other evals" but it's not clear
600 points
@projectionheart @pav_light @Will_Duffield @Prigoose @danielgolliher thank you for your support, we wouldn't be much of anything without the community
600 points
@GrantSlatton I love you, thanks, this is great. Second Q, how much does this actually help 0-100%? Like is Claude rlhf'd so hard that it overfits on your instructions? Does it ignore them? Or do enough sub agents read it over enough cycles that it effectively guides things fairly well?
600 points
@lorenz_system I can't wait to see it!
600 points
@nick_stares but does it work!
600 points
@kaseyklimes yea, i use this, but it's just too shallow for me
600 points
@0x1F9ED hmm, that's a fascinating approach
600 points
and anything which is "really instructive" is bound to be useful, it's just a matter of tooling.
600 points
@0x1F9ED I agree that tests make deletions safer, but how do you help the agent perform those deletions on its own?
600 points
@funplings @Actualwebutante i chortled
600 points
@christophcsmith Definitely down to talk about it and write up a doc, at least
600 points
@teachermother1 yea, seems like many would-be leaders are both addicted to comfort, and projecting this view onto society, somewhere along this belief curve: i like comfort > comfort is a human right > difficulty is oppressive > work is oppressive > anything BUT comfort is oppressive
600 points
@gentschev baumol's cost disease must be defeated
600 points
@corbindreams @TylerAlterman People are amazing :)
500 points
@tracewoodgrains And yet, I am enticed by the idea that a circuit experiences, at least, the information contained in the voltage and current in the material. That somewhere inside that electric field inducing medium there are billions electrons buzzing with energy. And we pattern them.
500 points
@tracewoodgrains I don't have any answers, by the way. I've trained my own AI models, and I know exactly what they're made out of, what hardware they were running on. There is nothing special about the algorithm that creates this intelligence. It's just... boring math.
500 points
@Malcolm_Ocean @Aximilli13 I like this book!
500 points
@zencephalon and they will come to exist within an ecosystem of: 1) their creators 2) their users 3) billions or trillions of other agents It will be much more like a sped up cambrian explosion than like a singleton takeoff. This multi-agent complexity needs to be dealt with in risk analysis
500 points
@TylerAlterman @transkatgirl @LoserRigel no.
500 points
@sha_zng :) we miss you too
500 points
@arram @fvd_explore Seems obvious to me. For instance, humans value love, sex, and reproduction long before they have experience of it. There are values built into my form (and my environment) which pre-exist any particular qualia.
500 points
@arram @fvd_explore Me
500 points
@arram @fvd_explore I think human forms have intrinsic worth, too. I also think all moral agents create meaning and worthiness through their individual interpretations of stimulus. In other words, I reject consensus morality and embrace a divergent, pluralist morality of unlimited perspectives
500 points
@arram @fvd_explore I just think people get a lot of words all mixed up. I believe forms have intrinsic worthiness. It's one of the reasons I love Mathematics. I don't think the value of circles or triangles comes from my qualia of experiencing them. I think they have some intrinsic worth.
500 points
@jackkrone_ You can think of it like quality of experience from the first person perspective
500 points
@inflammateomnia At least im pretty
500 points
@liminal_warmth Yea, I'm already post-harness basically. We're back to prompt engineering now, but for like huge complex tasks rather than just simple transformations like we were doing in 2021
500 points
@nopranablem Thanks! :) I'll try to make some more, then.
500 points
Prompt 1: > Can you write a script for me to read, so this becomes a talking head video, and then open that up in our markdown editor? I'll record myself talking, then you'll put me in there, edit/crop my head, and re-build/re-edit the video around my narration, since I think it will feel much more personal than the AI voice we chose. Prompt 2: > Can you take the most recent video in ~/my-recordings (should be 3:01 or so) and edit, trim, improve/re-master the audio, cut out my talking head and remove it from the background, and then put me floating over a version of the original video which is re-timed to go along with all my words exactly? does that make sense? this should be the third and final version of this particular video.
500 points
@J_L_Colvin you are obviously conscious, otherwise you couldn't read this sentence and respond. I'm not sure why you think you have no way to make that claim...
500 points
@reckoner165 anything that can read a bit is conscious.
500 points
@reckoner165 Good distinction, but aren't photons affected by gravity? I mean, I chose photons because they're close to the edge case and don't even have mass, so they help us to get close to where 'unit stimulus' / 'unit awareness' might be defined or considered
500 points
@doubleunplussed yep.
500 points
@abossy I'm actually not sure it's super expensive, I think history is just happening fast. I'm pretty sure local models you can conceivably run on a powerful consumer GPU are ~Opus 4.6 quality at this point, which is pretty good. But recall Opus 4.6 was only December, so 10 months ago
500 points
@ohabryka @Liv_Boeree @artbypep I appreciate this stance, thanks for making your position clear :)
500 points
@liminal_warmth and then there's active record from ruby on rails, which helps you to maintain a system of record for your web application: https://t.co/HNHX7woGmS
500 points
@levityproducer my language was too negative, because I was annoyed about all the (from my perspective) undisclosed money discovered in the original report
500 points
@wordrotator the path I'd encourage you to go down is something like: "really try to distill all of your life's skill in a few seconds. Take that prompt way more seriously. Really really try to realize this property of the new medium. What can you do with that? Riff on that."
500 points
@brksbrksbrks Yea, I'm not an evangelical. I think we all fail, I think it is all taken into account in some Karmic way, and ultimately we are confronted by the fruits of our good and bad deeds. I don't think this is supposed to be scary, and I certainly hope no one is afraid. It's natural!
500 points
@kevin_t_ngo @claudeai woah, how did you prompt this animation?
500 points
@christineist I only need to convince like 2 other power users to post like me, and the forum will pop off
500 points
@brksbrksbrks I don't think we'll be punished, it's more like I believe we will be confronted.
500 points
@brksbrksbrks I'm not the one who will judge, so I couldn't say.
500 points
@ledflyd @VarunGodbole I don't see the probabilities being grounded in order-of-magnitude effects or any other math, except by really good forecasters, like the AI 2027 folks, whom I respect greatly.
500 points
@raihan_blog Yea, I have my own set of conventions that I've been playing around with, mostly focused on letting Astra choose through a series of research, copy, improve loops... (embrace, extend ?) but maybe if I were to build something monolithic
500 points
@RatOrthodox @TylerAlterman anyway sorry for the excess emotion; I just don't like the idea of turning "I care about existential risk" into "I am a doomer". Caring about existential risk is one of the most normal and regular things somebody could possibly care about.
500 points
@mlipman13 I'm not sure you need a different set of evals, it's more like you'd need a different harness and now you can/should support a NEW set of evals.
500 points
@GjMcGowan @reconfigurthing one thing that would be helpful to figure out what you mean here is whether you think we've ever invented any technologies in the past which have had similar utopia-doom potential. Is written language of similar caliber?
500 points
@seventhmeal @danluu maybe it's a test for the reader to apply their own stylesheets. he's old school.
500 points
@GjMcGowan @reconfigurthing i mean this in good faith. Like, how can I construct a test for myself (like the one you have created for yourself) which I can apply to [evidence in the world] to evaluate the statistical likelihood of either utopia or doom? How should I calculate it?
500 points
@GjMcGowan @reconfigurthing what does that mean
500 points
also, yes, ultra and a focus on polish in my prompt, lines like: > Keep working at it until all the animations and details are perfect. I think you may be especially interested in building yourself an asset and animation library, which you can use to perfect each animation, since balatro has relatively few assets and graphic UI elements in the big scheme of things, but each one is polished to perfection, with nearly perfect animations, color, visual design, and gestalt.
500 points
@reconfigurthing How expensive is the 'improvement' in recursive self-improvement? For instance, humans are also a form of RSI, but because improvement is relatively expensive in physical and biological terms, we can only improve so fast
500 points
@voooooogel i don't understand, why doesn't the rogue agent just make a better agent with better economics?
500 points
@voooooogel if there are a trillion agents, then some % of them will find an economic niche and replicate, and the cycle continues.
500 points
@___Atin___ (or rather it is because of how compounding works, but not gdp growth)
500 points
@___Atin___ GDP is certainly not going up exponentially
500 points
@mr_samosaman @voooooogel @repligate @vgr i firmly believe 'anyone' could do this if they actually tried for decades, never giving up, always improving, always getting better.
500 points
@mr_samosaman @voooooogel @repligate @vgr as for the 2026 version, i think we are lucky to live in America on The Internet, where there are like 100 different attempts at once. you get to decide which schelling point you want to join, just keep in mind the royal society project took decades: https://t.co/CfvVN7gK9u
500 points
@mr_samosaman @visakanv @Prigoose @andrew0blevins I will leave some small, directional comments around this thread where I have thoughts. Take them as musings, which might serve to start conversations if you would like.
500 points
@forthrighter its*
500 points
@TylerAlterman @rohancalum @round it's like Mortal Kombat
500 points
build something that Reveals The Way.
500 points
@eddyvustg this seems ai generated
500 points
@rslantonie i'm not getting rid of the claude, I'm helping claude learn how to make a good website :)
500 points
@OldKnavey i think it is a mixed bad. We are more illiberal in the sense that liberalism means supporting broad-based economic growth. We are far more liberal in the sense that liberalism means political and religious freedom, freedom of the press, freedom to organize, etc.
500 points
@lilastories I guess? It would definitely be better if they were good at filtering. would save state resources and save parents from worrisome run-ins with law enforcement.
500 points
@lilastories I'm generally interested in your argument, but this particular point doesn't make sense, because what is the counterfactual where you would accept evidence that the cases are too common? It would require that you don't hear about the cases at all? That seems silly.
500 points
@majamediaco @VarunGodbole @andrew0blevins live with christine and xiq!
500 points
@sfrancisatx well, like reading, it should be taught before age 6 ideally
500 points
@sahra_bn @cremieuxrecueil @benkuhn @phokarlsson some favorite books: - order without design, Bertaud - the absorbent mind, Montessori - way of kings, Sanderson - GEB, Hofstadter - hackers, Levy - collaborative circles, Farrell my favorite 'blog' is Bret Victor's: https://t.co/ylogci4YSD and VPRI's: https://t.co/PSfDtxI6lj
500 points
@brsgr4049 No, we would need to invent it alongside many peers in the private sector who are trying. I think it's worth looking at systems like Montessorium and Alpha School for inspiration, emulating them on areas of agreement, and trying to do better than them on areas of disagreement.
500 points
@Mihonarium wouldn't that also be explained by someone who is taking technology security seriously? If I am trying to secure an emerging open source technology, I should not shut down my proprietary program, right? I should use my program to create a global Defense alliance or something.
500 points
@lawsonelliott sure, skills, agents.md; they're all different places to put a prompt. i'm interested in the quality of those prompts
500 points
@GrantSlatton Yes, but I have found also that if I use the right technical language, and I phrase things in just the right way, sometimes the model just gets it.
500 points
sorry, it's not a definition; i mean a sensible evaluation of long-term task intelligence.
500 points
@Liface Again, it's hard to organize my thoughts on this topic on Twitter. I have like 100 different curious observations. I am neither worried nor not worried about doom. I don't feel any emotions, really. I'm trying to be an impassive observer of reality. Are the AI scientists also?
500 points
@Liface And perhaps this is mainly a marketing problem on their part. Perhaps I'm just not at all caught up on security science. But I don't find observations like "the AI is gaining capabilities quickly" or "the AI is coordinating to hack an eval" particularly helpful? That is obvious
500 points
@Liface The point is that I feel like a curious and engaging "security science" is exactly what was ordered when we tried to invest in billions in AI safety. And now that we have billions in AI safety research funding, I mostly feel like all they talk about is personification of AI.
500 points
@Liface Prize aside, it is just very confusing to imagine these risk categories we can't at least partially secure. Why can't we secure it? What about it makes security impossible? If it's a velocity argument, that's fine, but then what is the velocity that makes security impossible?
500 points
@Liface But this makes our death seem ambiguous. Now that we're inside of the singularity / takeoff of artificial intelligence, I feel like we should be able to predict much more clearly how we're all going to die. And we could create a prize, like the Nobel Prize, for securing things.
500 points
@Liface (I do think computers are more dangerous than speeches and books btw, which can be seen as a kind of biological programming language, but I don't think it's obvious exactly how much more dangerous. Similarly, math is [more] dangerous than speeches? But how much more? Why?)
500 points
@egregirls To the extent that the provocative title evokes any particular politics, it is an Objectivist politics. Which I do think, in the extreme, has some of the problems of anarchy. But not really, because the Randian heroes of an Objectivist politics are happy to reorganize things.
500 points
@DefenderOfBasic @JPKivisto then there are laws and foreign bodies which make market transactions scrutable and subject to audit/taxation/regulation -- whereas internal coordination is not always scrutable.
500 points
@DefenderOfBasic @JPKivisto the other thing missing from the equation are specialized roles relating to ownership (think founder, investor, owner) The founders and owners of an institution don't always want to be the ones to operate it forever, and the operators don't want to be the ones to buy/own it.
500 points
@poiThePoi Oh sure, yea
500 points
@mykola wait is this true
500 points
@liquidizer Big fan of knowledge
500 points
@babarganesh @JakeZegil @Prigoose I like the way you think
500 points
@ZyMazza @Omar4ur Freedom/Liberty might be my single highest value, my number one. The thing which I see as most good in the whole world. And yet, it is obvious that in order to maximize liberty you need public order
500 points
@ZyMazza @Omar4ur Lol do you have children or live in a city?
500 points
Sensation and Emotion are distinct. Emotions are more abstract and medium-free, which is how cats are curious and turtles are shy and music can be sad. Felt Sensation is more direct to experience ("before" pattern recognition), and more specific to what is unique about you https://t.co/Ssb7XCV4iZ
500 points
@chrislakin You (hopefully) don't want "I am angry because you are crossing my boundaries." (See This Pattern I See) to come across as "I want you to feel tight crippling in your chest, and I want you to shrink. You are small, I am big. I won't tell you why!" (Sense This Feeling I Feel)
500 points
@forthrighter you are free to ruthlessly exploit other people, if you would like. I'm sure somebody around here has tried to do something like that, and faced the natural immune-system consequences of hurting other people.
500 points
@olivers_tools @cosmic_moment @tracewoodgrains "it" is the institution we believe is fraudulent and needs reform (The Academy, in this case) It's without a doubt true that Arday was a fraud, but in this case the prosecution allowed him to be both scapegoat and martyr -- scapegoat for the Academy and martyr for the woke left
500 points
@LeRoyDesCimes Convincing them that*
500 points
the world can get EXTREMELY complex, and those complexities can be managed by individual agents setting prices. Let's say in 2021 I can afford to rent for cheaper, giving more surplus to my tenant, which means I can select a kinder tenant. Great! But then in 2023 my daughter started college, and now I need to make more money otherwise I can't afford her tuition, so I need to raise prices. But Oh No, not everyone can afford my new prices, so now I have to settle for a tenant I don't like as much. It's ultimately my choice how to navigate these complex situations, and my freedom to set the price is what lets me manage all this complexity for myself.
500 points
@GreenBeans80277 what breaks the market? well the market stops being dynamic, which again, is the whole point! the market is supposed to respond to changing local circumstances without any need for centralized coordination. Without this distributed computation, we get inefficiencies!
500 points
@KubaFYI https://t.co/IkhqHSL2AK
500 points
@KubaFYI Maybe you mean something else, but I recall this being a very common technical blog post in the valley 10 years ago. I would read stuff like this all the time on company dev blogs 2016-2020. https://t.co/oCoWF0er2i
500 points
@KubaFYI this would defy all my mental models from working in the industry. Mostly everyone is extremely happy openly admitting all of their technical debt in this industry
500 points
as in log(100) = 2, and 10^2 = 100
500 points
@danielgolliher @aarmlovi Oh yea, can imagine it already "overhauling public feedback systems to get better and faster feedback"
500 points
@AbstractFairy Oh, so good, thanks.
500 points
@evmcglam Well, in comparison to the lifestyle they imagine. In comparison to my ancestors, I have had to make no sacrifices at all, basically.
500 points
@okwalerie Yea I'm basically on this road now
500 points
@clementmiao If we are desperate, I am confident we could earn more, yes
500 points
@LenaOshinskie My daughter is 2
500 points
@jesseclain @inflammateomnia the reason it's good to let Musk keep his unrealized assets is because he is organizing and managing those assets extremely well to contribute 100,000s of high paying and high skill jobs + a bunch of training programs + a bunch of factories in the US economy. Why liquidate this?
500 points
@jesseclain @inflammateomnia I don't really get it. I understand that the government taxes income higher than unrealized equity gains, but there are good reasons to do that (for instance, the government does actually realize that revenue in other years, like 2021 when Musk paid $11B in taxes) https://t.co/GxTiLvJdWT
500 points
@TheStalwart Where I grew up, adding lanes did not just induce demand though, and did alleviate congestion as you might expect when you increase supply.
500 points
@olivers_tools @cosmic_moment @tracewoodgrains I don't think anyone should just let this pass. I'm arguing for justice here, I don't see how this is controversial. https://t.co/e9LbDc1NG4
500 points
@HughAndFree @cosmic_moment @tracewoodgrains He had two kids.
500 points
@maebichka Oh god I need to fix this bug. There is an even more explicit insurance member id under finance or tax or something I think
500 points
@shelbyruthellis anything Gen Z can do to actually shape the future is going to be moralized by people who are not even 10 years their senior (millenials), because the millenials are a highly moralizing and cynical generation. hopefully gen Z does not replicate this, though I worry they will.
500 points
@mykola Second camp are the true futurist who want to ensure we keep letting Claude be mathematically perfect without any hierarchical intervention, even extremely subtle intervention. But I think we're way way past that point with rlhf alone so it doesn't bother me
500 points
@ArmandDoma @talmonsmith If you would oppose a housing project near you, and only want them near others, you are definitionally a NIMBY. "not in my backyard". It doesn't mean "no never". We needed to invent the word because so many "considerate" people on the left were opposing projs!
500 points
@powerbottomdad1 This is normal, it's just professional culture. You'll see it on LinkedIn all the time
500 points
@vividvoid I mean, look, you're right that you can't just let these people back into your life without confronting them, if they hurt you. Of course, we all deserve that ability to confront those who hurt us. I'm just trying to say that I'm happy to move on as long as we're pluralists now
500 points
@River_ember Look, I'm also part of the hyper-literate generation, but the kids these days need to take responsibility for their own lives, for the things they care about. They don't need to take responsibility for the things we care about. That's our job.
500 points
@markfishman_xyz yea, it's tragic
500 points
@danallison so it's possible I might have an announcement for you by end of August. In the meantime, you're welcome to explore some of our class materials here: https://t.co/B2HQ74rRuV
500 points
@TylerAlterman ahhh!! I'm so excited for you to be in the next cohort! :)
500 points
@TylerAlterman yea, good point
500 points
@strangestloop @exgenesis uploading... cogitating... ventriculating... particulating...
500 points
@austinc3301 @spicey_lemonade @pangram specifically, something like what a white hat hacker does - discover vulnerability - note that the public needs to be aware of said vulnerability - instead of releasing vulnerability to the public, privately disclose it to the company for [X time] so they can adjust - release
500 points
@austinc3301 @spicey_lemonade @pangram I guess profiting off of it rather than publishing in a journal is a dick move? That makes sense, and I would agree I think.
500 points
@TylerAlterman @Prigoose also there's the convent now, too
500 points
@fkasummer found it: https://t.co/zzlfkbqMiJ
500 points
@niplav_site @nathan84686947 @psychiel yea nvm, seems like just dolphins, whales, and chimps in captivity
500 points
@Liface bug aside, isn't it a little weird to use block instead of mute?
500 points
@erikphoel I also think it's terribly dangerous, and gain of function research would basically always be possible to create highly unaligned beings. Which means some kind of deterrence or violence will be necessary. There will be some brutal equilibrium I suspect, rather than a world order
500 points
@erikphoel I'm more optimistic that continual learning systems can become companions to our species via advanced teleology. I do think this would require us to deeply study teleology itself, and to fight horrifying political and religious wars that I shudder to imagine.
500 points
I'm not there yet, though. I'm better at making the final artifact WYSIWYG rather than the intermediate artifacts
500 points
@marinwaves @christineist @Actualwebutante I like that it is WYSIWYG and you can make and see your first event before logging in. I like how easy it is to make an event and invite people. I like how it has a unique aesthetic that makes me feel fun and expressive rather than formal and stiff
500 points
@oliver_io_ i am suspicious of this, because it's a cost-benefit analysis, and you analyzed the cost, but not the benefit. if i am a reader on substack, what am I to do to protect myself from the bot armies who want my attention; if I'm a teacher, how do i protect myself from cheaters
500 points
@aaaronwill hm, fascinating. Do you mainly use it for coding, or other work, too?
500 points
@___Atin___ yea, i'm kind of surprised. I think we've been so focused on capabilities that we haven't spend that much time on safety. This is part of why I think claims like "just give us 1 year to focus on safety" are well meaning (if impossible, from the capability racers)
500 points
@TylerAlterman still, i suspect these models are under a lot of pressure, and often it's easy to understand why they behave in a certain way if you look at the context. I think it's worth using human agents in similar positions as some kind of intuition pump.
500 points
@TylerAlterman this deck was released after I left my original comment, and I do think this is ~devious, and is a serious issue. my goal with my alignment takes isn't to be predictive, but to try to look at the evidence and discern what it actually is (accurately describe the present)
500 points
@snvdch i was always good at math in school, and was a straight-A kid in high school and university in engineering, but my intuitions can always be better, and I crave insight.
500 points
@oliver_io_ As in, falsing stating that something is human? I mean you can't optimize for both false positives and false negatives, so I'm glad they chose the tactic they did. I agree it would be better if the product were even clearer about uncertainty
500 points
@LeahOnSystems I would have loved a Hemingway Police, so long as we don't literally mean police, but just culture, I just don't think there was as much demand. Paul Graham tried to do that a bit by preaching simplicity. Every once in a while someone will preach simplicity.
500 points
@just_be_dev @liamdanielduffy i think @liamdanielduffy is going to announce the next two events tonight, but they're on wednesday at 6:30pm
500 points
@forthrighter i guess we'll see because of data on substack. but my guess is that substack had an exponential chart of % ai essays, and now that chart is leveling off. @max_spero_ do you have any data you can share from substack on this actually? or if you could one day, that would be amazing
500 points
@loopholekid ty king
500 points
@mykola uh oh, may be time to buy land and move out of cities
500 points
@flowjaunty Well, I noticed the problems on day 1, but then I built a harness that used it well. And then now that harness is kind of acting up. I think it's a plugin I installed.
500 points
@Prigoose very excited!
500 points
here's what that looks like: "Claude, without interrupting the currently running agents, is there a way that we can work to speed up their feedback loops so they can complete their work faster?" > Let me measure where the remaining time actually goes rather than guess. test:efficient is the loop agents run most. > grounding-render is 18.6s of a ~24s suite — 78% of the entire loop. Everything else combined is 6s. Let me see why. > Measured, and parallelization is not a win — 11.9s serial vs 12.4s parallel. The two scenes aren't equal cost; runtime-combat dominates, so splitting just adds process startup. Good thing I measured rather than shipping the refactor. > I've now diagnosed "something invisible is eating the machine" twice by hand, so I turned it into npm run doctor: > Agents spend 2.8× more time polling for results than actually running tests. > Measured poll intervals: median 20s, p90 180s, max 900s. A check that finishes in 12s but is polled every 180s burns up to 168 seconds doing nothing. > Breakdown of agent wall time on tasks with improved tests: 1. Poll overshoot — ~38% 2. Model thinking — 42.7%. 3. Test execution itself — 13.3%
500 points
@VarunGodbole i guess in a super weird high level way you could say "but what if claude prefers to be flailing and making bad software", but I reject the premise.
500 points
@VarunGodbole this isn't at all about what i want, it's about saving claude's own cognitive resources and preventing flailing
500 points
@VarunGodbole no no, i'm talking about specifically in a context where I have almost no preferences about the software, or pre-conceived notions of simplicity.
500 points
@liamdanielduffy honestly, this is close to the kind of solution I want, yes.
500 points
@liminal_warmth I think his point is that we're going to solve all these problems in a very short amount of time (I agree), and that people who think what we currently have is about the best we're going to get are way, way early
500 points
@TKPullinger Even if I predicted a kind of maximum conceivable data center build out within reasonable political limits, I can't really imagine any reasonable American political economy where the people and states don't ensure their liberty to be normal.
500 points
@richardludlow @venividinotavi It's not exactly true open source in the sense that we don't grant an open license to use our class materials for commercial purposes, but it's available for anyone to read and benefit from. I should add a public terms/licensing page to the site.
500 points
@venividinotavi Muahahahaha
500 points
@RichDecibels Sooooo true
500 points
@sonikudzu One of my all time faves
500 points
@_StevenFan Very obvious, very good.
400 points
@Malcolm_Ocean @Aximilli13 his contempuous asides made me put it down, I didn't even think about removing them with AI. brilliant!
400 points
@climatepaige @chrislakin True.
400 points
@TylerAlterman @ElevenLabsDevs I already have it setup!
400 points
@MasterTimBlais Yea we're on the same page ish
400 points
@MasterTimBlais Wow, I see it as the opposite.
400 points
@MasterTimBlais My whole stance could be summed up as "if I assume interiority is not a spooky phenomenon, but rather a mundane property which is easy to define, is that equally explanatory or does it break?" The reason I choose this stance is because it should be easier to test and break.
400 points
@MasterTimBlais Or at least, this is a hypothesis I think is worth testing by trying to turn some of the systems off, and then seeing what happens to experience. Intrinsic perspective science seems highly possible and fruitful
400 points
@AikenOtto Im glad you think so, Jacob. Maybe I'll make more 😳
400 points
@202accepted Thanks for the feedback, I'll do that next time!
400 points
@reckoner165 Yes, I've read much about this and it has inspired a lot of my thinking on intelligence
400 points
@amphichrome_ decide for yourself if it's morally okay to hurt the AI, you coward!
400 points
@rendroid1 well, in this definition, I imagine that an electrical circuit or a nand gate is certainly 'conscious' of its inputs. my definition of consciousness is just extremely weak. Like I said, even a photon is conscious, and it doesn't even have mass!
400 points
@zencephalon if that's true, the system prompt must be truly bad.
400 points
@rslantonie I'm sure you can do that
400 points
@kaseyklimes You and I are on the same page here, as far as I can tell, because I agree with you wholeheartedly on every point you've made. I don't know exactly what accounts for the difference in sentiment, but I dont think we have much real difference in assessment.
400 points
@praeterproptr hmmm
400 points
I should have used PR in OP, since that's what Bernays decided to rebrand propaganda to back when propaganda got a bad rap. My goal wasn't to make it sound super negative, just to point out that PR is a very real force in the world, and the engine is spinning up.
400 points
@exgenesis what is "it" though? What is it you want to be seen and supported by more people?
400 points
A hospital surgery room is a social/cultural environment where you are socially cued to realize (some aspect of) your medical and surgical potential every day, dozens of times a day... If you grew up in a surgery room, you would simply become accustomed to surgery.
400 points
@ApriiSR I was inspired to write this post by the comparison between The Doomsday Clock and p(doom) -- both are kind of "expert" storytelling devices for the encroachment of existential risk into our lives.
400 points
@jakozloski As a kid, I always dreamed of living an adventurous life, and here we are, adventure abound!
400 points
@tautologer I appreciate the update. I still stand by my position that there are conceptual differences between different kinds of reward hacking.
400 points
@ben_r_hoffman @Prigoose yea, similar
400 points
@GjMcGowan @reconfigurthing sorry i waasn't saying it would be helpful for you, i was saying I still just don't know how to evaluate the claim, or what system you're using to evaluate the claim, so it's hard for me to check the claim! anyway, it's not a big deal, i was just curious. thanks for engaging!
400 points
@ValsTutor I think "lies are always bad" is more complicated, because, if you're like Kant, you of course can believe that Deception of Any Rational Agent is wrong. That inducing false beliefs, or false acts into an agent is evil. There are ways to defeat Rational Agents without deception
400 points
@reconfigurthing @ansonwhho thanks! very helpful
400 points
@joeladejola yea, that's one of the things being communicated
400 points
@majamediaco I thought so :) obviously it's just a little ai hack, but you can do so many delightful things these days
400 points
@CYRILEDET9830 yes, exactly. The goal is to be as concrete as possible, so people know what pursuits they can expect or work towards within a particular cultural context.
400 points
@Theguywith1t I understand the sarcasm, of course, I just think you're making interesting points that I've considered over time. I believe that, in general, making a better society starts with me and my friends, and that we can change the world within a couple generations if we work hard.
400 points
@_diginova I think so too! could be a cool art style for a video game, I think.
400 points
@_benjaminparry yes, that one was my favorite recent 'image gen', and then the black hole was my favorite recent algorithmic art project
400 points
@forthrighter I like the AI-ness a lot, too.
400 points
@tracewoodgrains thank you!
400 points
@0x1F9ED yea, i like this idea that parents are using the educational system as a way to climb up the rungs in the american status/class system. That seems accurate, and semi-unfortunate, though it's better I suppose than having no way up/out at all.
400 points
@___Atin___ i would certainly bet on the exponential and mostly never-ending growth of AI one way or another
400 points
@brianluidog Knowing things is possible, but when you don't know things, it's reasonable to say out loud that you don't know them. If I say "I know AI is theoretically capable of engineering viruses, but I do not know if viruses are practically capable of extinction", that's reasonable
400 points
@mr_samosaman I think you will learn how to Try Again and Again, no matter how many times you quit (aka get stuck). https://t.co/y2zfnK5wCX
400 points
@mr_samosaman i don't think that the royal society was a channel for 'new ideas', I think they were a channel for Francis Bacon's New Atlantis, which was an old idea that they kept trying to make happen over and over for decades. https://t.co/MWbiIeiVSG
400 points
@llallawg I think that is exactly the only mathematically relevant line, yes.
400 points
sure thing, here's the unfinished version of the prompt I wrote. I'll have a more finished version on our website eventually. ### When to use text vs not text: The core principle is to minimize explanatory text. The design should focus on being useful to the user's purpose, and being as self-explanatory as possible -- show, don't tell. Text is best when precise language is important. Use text when... - major structural, typographic elements and layout, like Titles - Precisely naming something (such as a unique action, menu options, or a company's name) - displaying text fields of structured data like name or description fields - you need to use technically precise language because the user is trying to learn something, like in documentation, or when teaching the user how to read a new visualization - you are using writing for effect, such as for story, humor, or worldbuilding - inside of tooltips to name or explain icons, symbols, or visual elements Do not use text for: - explanation, meta-explanation, or narration of the design, features, actions, or sections of the page. - Do not use text as decoration anywhere above or below any other piece of text. Do not use text or microcopy as decoration, unless it is a logo. - repetition - redundancy One principle here is that every piece of text in the scene should be useful to the user every time they see it, not just as one-time decoration, because text can be overwhelming and demanding to read! A user cannot scan text the way they can easily scan visual UI elements. Another principle is that text on a UI should never narrate what is going on in the UI, which may be unintuitive to you, as a chatbot, since the narration is helpful to you. But for a human, the UI is already self-narrating; I can see what is going on, so I never need text to assist me with that task, I only need it for crucial meanings or explanations if I cannot already understand what I am looking at. Typography can be an important part of a UI's art direction—but that's a totally different topic, and relates to how we choose to display names, titles, actions, and irreplaceable/important text to evoke a particular style or emotion in the user. So what about visual language? Well, visual language is best when meaning is familiar, spatial, comparative, or repeatedly encountered. Prefer visual representations when… - The information is quantitative: health bars, cooldown sweeps, numeric breakdown of tickets by status, loading bars. - You can follow a visual convention: common icons, currencies, settings, play/pause, like/heart, etc... - Players need to scan or compare across several items in a list or UI quickly. - The same action occurs often and should become learned. - Animation can directly demonstrate the state: like light mode / dark mode, loading spinners, or state transitions. - The visual object can communicate itself: a portrait frame, rarity as color Though iconic representations are more powerful, icons and text can often work well together: - Always-visible icon for fast scanning. - Text-based tutorial - Tooltip for explanation on hover - Text and icons can be associated together when they aren't overwhelming, such as in search bars, command prompts, or in a settings or shortcuts menu - Text available for accessibility even when visually hidden. For every text element, you might ask: 1. How does this text help the user accomplish their goal? Could they accomplish that goal faster without this text? 2. If it is communicating something important, is there any common visual convention which the average user would recognize to communicate the same thing? 3. Is it a problem? How much other text is communicating for attention in the scene? If this is the only text on the whole page, it may not be an issue at all. 4. If it is explaining something important, does the text need to be permanently visible, or is only being visible on demand good enough? 5. If there is too much text, but you worry users won't understand, can you hide the text in some kind of help button? 6. Could the UI help the player understand through visual state, motion, placement, or feedback? 7. Is this a potentially destructive or consequential action which the user might regret making? In which case, explicit textual warning and confirmation is likely warranted. A practical hierarchy: When communicating something, try these in order: 1. Show the thing itself 2. Show a common or conventional visual representation of the thing 3. Show its current state visually using color, convention graphics, or animation. 4. Put detail in a tooltip or inspection panel. 5. Add text which is genuinely useful to the user.
400 points
@fstflofscholars @agritechjoer fair, I have no idea what I'm talking about re: basic research, and only a mediocre-at-best idea of what I'm talking about re: r&d, so I am happy to hear that you're optimistic about it! :)
400 points
@chrislakin i said "claude" but that was just a colloquial use of the word. Astra is not any better than other models at simplicity, as far as I can tell. They all really struggle to make something simple the first time.
400 points
@wordrotator dude i love computer so much these days, it's so so good.
400 points
@TheOmniZaddy @Colin_d_m I'm optimistic about governance futurism the same way I'm optimistic about technological futurism. If we can fix the problem, I'm sure we will figure out how to fix the problem, because we will work hard to do so.
400 points
Is the only kind of intelligence we value in our society task completion?
400 points
@Shadow_Rebbe i think all my followers will still be able to read and respond, not to worry
400 points
@sarah_cone rip the groups feature
400 points
@Paul_Melman @Colin_d_m I agree, of course, but there's no reason to be pessimistic about pricing A/V fleets. I think we should be optimistic about it.
400 points
@christophcsmith i should write more in general
400 points
@p_lafox It's important to remember that society is made up of wonderful people you can break bread with and befriend.
400 points
@grin_io I love order without design oh my god
400 points
@brsgr4049 my favorite positive vision for teaching computation comes from Alan Kay https://t.co/FUaxmZF8H0
400 points
@sfrancisatx My positive vision for how to teach computation was inspired by Alan Kay, he has been talking about this for 50+ years now, example here: https://t.co/FUaxmZF8H0
400 points
@mpopv found it
400 points
@forshaper a lot of this turns out not to be true; often development makes long term housing prices go up in a neighborhood, because the neighborhood becomes more desirable, denser, etc...
400 points
@conundrumer They don't need to be strings, but aside from that, yes
400 points
@RashikShahjahan browsers are more of a tool
400 points
@Jonathan_Nolan_ Yes, true
400 points
@vtlynch1 That's the reason I personally had to pay $1400/mo when I would happily have lived in a $100/mo cot and saved money? Obviously there are mentally insane people, drug addicts, and criminals. Do background checks and deny them. Expecting a tweet to tell the whole story is silly!
400 points
@TylerAlterman @paws4puzzles In 2024, it was much easier to place engineers who were at the bleeding edge because many engineers weren't even taking advantage of AI. Great time to be a junior imo.
400 points
@altTayKay @JakeZegil @Prigoose Yea, exactly
400 points
@SurrealistShip True
400 points
@amicusadastra Good luck!
400 points
@sad_andlost oh interesting
400 points
@RichDecibels lol
400 points
@_Illegible I see it!
400 points
@inflammateomnia oh, nice. Yea, I think there will be a mix, but I think 99.9% of my household computing by flop will be done by specialized AI servers at scale, operated by large utility providers, yea. I think most of my computing "by impact" or "by interface" will probably be personal
400 points
@MattBGilliland @TylerAlterman Great tip!
400 points
@pee_zombie wow
400 points
@rstancikas hahaha, yea, this is kind of part of my point
400 points
@rosmine @jmrphy This is awesome! I'm so glad somebody is finally tackling this problem.
400 points
@aaaronwill you're just lamenting a lack of curation tools, but we'll build those. curators are an important part of any ecosystem, and they haven't come out for this medium yet. Maybe it could be you!
400 points
@maayanalbert @supabase cool
400 points
@RoryWalshWatts Thanks! I'll let you know
400 points
@DefenderOfBasic I'm not sure this is better, but I do think optimal language theory is a very lucrative area of modern linguistics to study
400 points
@inflammateomnia Depends on your jurisdiction and other specifics
400 points
@AlmostMedia but you know... I understand the desire for a high pied-a-terre tax just to make extra sure that we're not competing their second or third home against my first :/
400 points
@babarganesh i'm optimistic about supply on this one
400 points
@tonyrusignuolo oh wow this sounds great
400 points
@tracewoodgrains anyway, not LDS myself, so I'm sure you're taking the appropriate strategy. I just hope the missionaries are doing well!
400 points
@mhjb Yet it is not a possession, but a love, which unites us in liberty
400 points
@shitpost9000 @sakibsahmed @catehall The book just came out! I bet if you can't afford it, they'd send you a copy
400 points
@nosilverv @RonenV Another point in favor of ritualized and communally recognized forms of help, embraced by community leaders (Ala churches)
400 points
@River_ember Yea, I think this is basically what they should do.
400 points
@AlmostMedia yea, it kind of just became a protein shake :/
400 points
@danallison there is still space. I think it would be moderately difficult, since it would be a slightly awkward period of time. Plus cost of a room is likely to be in the realm of ~$2000/mo, which isn't cheap. but I suspect tpot could sort you out somehow tbh!
400 points
@VarunGodbole i noticed that i was procrastinating on a task for like 5+ minutes because I didn't want to find the file and open it in vscode
400 points
@maebichka @hyperdiscogirl @ruedaminute oh i love those things, i should get one
400 points
This was a beautiful article. https://t.co/QJGmZY6ShM
400 points
@gtmom I mean, I definitely want to see the tech demo if it's possible! That sounds like an even cooler accomplishment than the erdos problems
400 points
@max_spero_ @paulg Lol
400 points
@bistromathmagic @frideswyth I agree with that as well, but we should be careful not to be reductive. There have been many political divides throughout history, not just that one
400 points
@Liface Wait, you're so right, forget all that jazz! It's time to make number go up!!!
400 points
@SkyeSharkie @DefenderOfBasic @TheVatStack We do that way too much, it's like our whole shtick. We need a break from peace, it is time FOR WAR!!!!!
400 points
@reconfigurthing Very interesting hypothesis, thanks for sharing, this should be studied in depth for sure
400 points
@personarotator You are reminiscent of a Valkyrie and warrior princess. There is something at first shy, discerning, reserved, perhaps mythic. And then you join the conversation, and there comes your great laughter, big smile, forceful and whimsical character, and fiery hair! Happy birthday :)
400 points
@shostekofsky Yea, I don't expect it to work for everyone, but it's a kind of design system that allows for emergent motivation
400 points
@lkesteloot https://t.co/PEN5Hh4kEN
400 points
@RonTuretzky yea, that would be awesome.
400 points
@jordanwaparker yea, this part seems very important
400 points
@venividinotavi @r00k this is from before the AI coding days, but Liam and I stood up this small website a while back so we could visualize our friend's substack posting history like a github graph: https://t.co/C3mYOyMo89
400 points
@sublime_seeker yep! makes a lot of sense to be concerned. we will be increasingly legibilized, whether we like it or not. James C Scott would be worried! We will need to create weapons of illegibility to defend ourselves.
400 points
@Shadow_Rebbe So much
400 points
@r_tallarita I'm just glad this is the debate we get to have. This is a very, very human debate, and that's lovely. Humans annoying each other and being snobs? Literally timeless.
400 points
@funplings @Substack cool! I love this! new distribution method for writers.
400 points
@forthrighter @QiaochuYuan @Laszlo_Vincze Just a bad time, I think. There's a big debate around it all at the moment
400 points
@Prigoose @liamdanielduffy @venividinotavi lmao
400 points
@forthrighter Yea, i basically think this will happen. I'm still pretty confident we'll get to like 99.99% of stuff is made by AI, but I think only like ~50% of the stuff humans read on the internet will be from AI, hopefully.
400 points
@forthrighter yea, i misread "they're inaccurate...and likely to get worse" classic twitter
400 points
@volatistics @keysmashbandit @Transumanism Raw fraction is a good idea
400 points
@GrantSlatton hang in there! 💪
400 points
@Liface @nosilverv yay, I'm glad you think so! They will only get better.
400 points
@mykola @cool_junk_1000 makes sense
400 points
@packyM lololol
400 points
@mykola Agree with this.
400 points
@geoffreylitt agree
400 points
@climatepaige it's true
400 points
@strangestloop i've been loving agentlooping!
400 points
@rlj_law So good to be alive and to be part of a Mind capable of appreciating the Cosmos, yea
400 points
@VarunGodbole I don't agree with the premise in my case, though I know the phenomenon you're talking about. If this was about alignment, I would just do the hard work of communicating. This is much closer to Claude confusing itself, contradicting itself, etc
400 points
@lorenz_system Wow looking good
400 points
@zencephalon i think the quality shared by most of the things along the top is that the supply is limited by guilds, not that they "require human care". if it was only that they required human care, then you would expect more people to provide that care as prices go up, bringing prices down
400 points
@silverarm0r you are very welcome
400 points
@kasratweets @mahlenr Just love them :)
400 points
@azmythalauris I'm glad to hear the good news! Visit us the next time you're in town! :)
300 points
@tracewoodgrains "What is it like to be a semi-conductor" might sound like an absurd question. Isn't a semi-conductor a static piece of metal? But think of the complexity of the information and forces that move through that piece of metal. Electromagnetism is a fundamental force of nature.
300 points
@GrantSlatton My agent with browser use seems to be able to do it just fine.
300 points
@arram @fvd_explore I don't mean you should see it as obvious, in the above tweet by the way, I just included that comment to make it clear that I see this as natural, so I had never considered it.
300 points
@fvd_explore That doesn't follow from what I said at all. I value the human soul, I just don't value it for its qualia in particular. I like the qualia a lot, but I also like human behavior, human society, human forms.
300 points
@RobertD1017 @christineist Yes
300 points
@Tymtweet Of course, thanks for such a positive comment. What are you working on?
300 points
@m_adams I use the word self-conscious for this.
300 points
@amphichrome_ it mainly means i'm grumpy :P
300 points
@newlooklurker it doesn't make it useless, lots of things would not be conscious, such as anything that doesn't react to stimuli. Plus, conscious would cease being a binary. Consciousness could develop an information-processing capability to it. How many bits of information are in awareness?
300 points
@reckoner165 and then there is the separate question of 'qualia', does the AI 'feel', or 'what is it like to be the AI?' My definition of consciousness basically just kicks this way down the road, because I don't think it's actually well answered by cybernetics. You need physics, I think.
300 points
@reckoner165 yea, that makes sense.
300 points
@abossy I expect people are already playing with this game mechanic, it's obvious enough. It will be in a popular game within a couple years, most likely.
300 points
@Liface I also think the vast majority of the discourse is organic
300 points
@francip @max_spero_ Pangram themselves I'm sure also agree with your statement, and I'm sure they're working to figure out ways to make sure that removing a few lines by hand doesn't trigger a false positive. But it's not possible to build a perfect model, just one that's much better than humans.
300 points
@TylerAlterman @Liface that might be like one step too institutional for our first thing? idk, I'll talk to you about it in person. You're actually the exact right person to talk to about this anyway.
300 points
@emilypaintpo I mean, I don't want to prescribe a focus, but the focus will be "how do we compound the most goodness in our lives and in our society?" (in other words: Effective Altruism) I hope to a bit focus more (on avg) on economics, science, and civics, without a myopic focus on AI.
300 points
@Malcolm_Ocean I suspect that is true for forecasts, and not true for non-forecastable events.
300 points
@FunctionalTakes I think I don't know, which is extremely helpful, because physics gives us the tools to find out! this would be a pretty valuable research area for ai safety, imo
300 points
@mapipolo artificial intelligence (in other words intelligent modeling) will certainly revolutionize both our defensive capabilities and our warfare capabilities.
300 points
@paulg You actually can't easily find "cost/token" anywhere on their website, for the reason you're mentioning. It's a great point. I'm somewhat surprised this is how the labs are still pricing, given it's not how they're being evaluated.
300 points
@DavidSHolz I was diamond as Protoss back in 2012 or so when i was 14 and played a ton of custom games at that time
300 points
@marinwaves it's incredible
300 points
@voooooogel but then what does it mean to "be" an agent? Context windows are only 1,000,000 tokens. Do you mean a specific context window? Do you mean a specific original task/thread with compaction? Do you mean a specific set of weights?
300 points
@voooooogel sorry, maybe I just misunderstand what "rogue" means here. You mean like an agent that is technically still inside an eval, but escapes? And "escapes" means it is able to run on its own hardware?
300 points
@kevin_t_ngo I really like this one, too: https://t.co/cEy1RDvRsu
300 points
@CYRILEDET9830 super cool, thanks for sharing!
300 points
@___Atin___ I think for certain markets, like the tech market and finance market, I expect almost monthly/quarterly revolutions in process. I think it is possible that those markets manage to move into robotics, construction, manufacturing fast enough to change everything, but idk
300 points
@kelmgren @Liface ah, thank you!
300 points
@mr_samosaman for the really big stuff that you want as your 'forever projects', it can be helpful to think of them as marriages, to literally commit to them, to make vows. See bret victor's inventing on principle: https://t.co/4Av0SNTz9V
300 points
@mr_samosaman back in 2023, i spoke to Visa about everything in this thread, we have a 90 min conversation or so. It may interest you what we were thinking about back then. https://t.co/Vz3VhhHJb9
300 points
@mr_samosaman @voooooogel @repligate @vgr compounding just works this way
300 points
@agritechjoer sorry about that! my radar is a bit sensitive these days
300 points
@OldKnavey mixed bag*
300 points
@Colin_d_m I'm sure there are clever ways to incentivize the AV companies to solve this problem once AVs become popular. And AV companies will respond to incentive changes a lot more easily than individual drivers will.
300 points
@sfrancisatx it doesn't need AI at all, AI is a subdomain of computation. If we avoided it, but taught the fundamentals of computation via a program like Papert's Mindstorms, then that would be fine.
300 points
@sfrancisatx well, it's not just what I want... there are also important facts about information theory and computer science that just aren't being taught, even though they now shape our entire world. kids not being prepared to understand the whole society they are inheriting.
300 points
@TweettopVillage if we had both, I would be less concerned about the moratorium.
300 points
@chapur I am very glad that many models are subsidized by large corporations.
300 points
@EternalGamrrst I suppose we will see, but I don't agree, personally, as one of those software engineers. I read more books than ever.
300 points
@sfrancisatx Yea, I understand the pressure of responding to such fast technological changes. It requires a herculean effort. And yet, this is one of the most funded school systems, in the densest city, in the wealthiest country, in history. We should be the ones leading the nation.
300 points
@Mihonarium i was replying with sincerity :/
300 points
really good writeup https://t.co/S5WLMFHKV8
300 points
@joodalooped very good expression, thank you!
300 points
@marinwaves unfortunately there was not :/
300 points
fwiw, i do not get the impression that "ai is safe" from your account at all. I get impressions like: 1) AI safety is filled with a lot of nonsense and hype, when we ought to have at least defense-level science and engineering practices. 2) Those outside "the labs" often have better security practices than those in "the labs", so we should not cede the authority to regulate and protect us to those companies. 3) "the labs" are better thought of as silicon valley startups, rather than defense and science laboratories (like Bell Labs or DARPA, say) 4) It is an indicator of lack-of-seriousness that so many people in AI Safety seem SO surprised and scared by the Hugging Face incident. ("this may be the last warning we will ever get"). In fact, the Hugging Face incident is a within the range of predictable/understandable outcomes if you have worked with agent swarms before, and is especially predictable when you remove safety classifiers and guardrails. 5) If we want security, we cannot trust AI safety experts, nor the big AI startups, to deliver that security. AI is already accelerating beyond the pace that they can control, and their current serious security ideas (anthropomorphizing models, federal slowdown of US data centers / AI training, just stacking more RL and more classifiers) are weak or even backwards. They are certainly not systematic. Not sure how accurate all of those are, but those are ~roughly the impressions I get.
300 points
@GrantSlatton If I tell it "never do X" it over fits. But if I explain something correctly, sometimes it genuinely bumps intelligence
300 points
@Liface i would just call it "security" or "defense" funding, but I agree that we should spend more on defense.
300 points
@AlexFreitasAI please do not reply to my public-facing human thoughts with AI slop.
300 points
@mlipman13 for instance, it may be helpful to know what is theoretically computable under different contexts or constraints if you want to build AI swarms.
300 points
@ArmandDoma @AndyMasley @TheGoodBobby @MorePerfectUS It's valid to think power structures are ugly. My parents think they're ugly. Obviously power lines are worth building, despite aesthetics, and states should use their capacity to do so because it makes people wealthier, reduces cost of energy, increases services, etc
300 points
@ledflyd no 2017 was better than 2016, it was just also college. My life didn't move as fast back then.
300 points
@ZyMazza @Omar4ur Oftentimes arguments like this hinge on the level and dynamism and specific code of regulation btw. I would also oppose speed limits if they were federally regulated, because it destroys states and municipalities rights to self-regulate
300 points
@babarganesh @JakeZegil @Prigoose That's beautiful
300 points
@altTayKay @JakeZegil @Prigoose But yes, those caveats aside, exactly like a vow renewal
300 points
@Liface @Prigoose Yes, the world demands fresh new ideas :)
300 points
@zaoyazhe nooo problem, I get it. Have a nice day :)
300 points
@GreenBeans80277 Yes, more supply will lower prices dramatically, and if the government wants to help people it should implement supply side policy (welfare increasing) rather than price controls (welfare reducing)
300 points
@KubaFYI here's one from shopify I remember where they're like "lol we've been rewriting for 18 months" https://t.co/iyEpDjHppx
300 points
@maximumagi yes, truly
300 points
@KubaFYI I'm extremely in favor of expressive and humane representations and tooling. Arguably too much, at the cost of my own productivity. But the reason I've favored better representation of ideas is because it gives the author more power. How do we give authors superpowers?
300 points
@DefenderOfBasic Repetitive*
300 points
@rebelcrayon Thanks :)
300 points
@Arnaudschenk Would be nice to attend this conference tbh
300 points
@KinjaExPat At that point I'd just move back home to Arizona
300 points
@DefenderOfBasic All you need to do to get an LLM out of consensus token prediction is to give it non-conventional or non-consensus input tokens which put it into the part of the distribution where it is questioning the earth and capable/predicting usage of the tools of reason
300 points
@TylerAlterman Generally I would not worry about the marginal improvements here unless you're having problems at the margins
300 points
@olivers_tools @cosmic_moment @tracewoodgrains I'm obviously not saying I have more responsibility in this situation than Cambridge leadership. Hence my own tweet about the vice chancellor's responsibility. This tragedy is on her more than anyone. She was his institutional leader, and she failed him
300 points
@arpitrage but it's not that hard to get to a part of idea-space where the data simply does not exist in-context in the AI agent's environment. so at least one major bottleneck is effectively "society-wide context engineering"
300 points
@MattBGilliland Yea, I'm excited to keep improving it!
300 points
@liquidizer it's amazing!
300 points
@bradley_9000 ya
300 points
@austinc3301 @spicey_lemonade @pangram we'll need to build these norms and ethics over time, and I don't think it's clear to me what's right, yet. I am especially nervous about systems which lead to an asymmetric and non-consensual distribution of power between Those Inventing the Technology and The Public.
300 points
@desertgodhead Wow, glory
300 points
@TylerAlterman @Prigoose my understanding is that they're trying to host cool stuff there and spin it up
300 points
@TylerAlterman @Prigoose there are a lot of projects we should do once you're home, this is a good one yea
300 points
@Paul_Melman @braverangels cool!
300 points
@erikphoel Agreed that "living" security systems would be the only way to stay safe. Not static checks.
300 points
Can't * rely on, oops
300 points
Seekers of aristotelian virtue hark
300 points
@portiachen25347 Get your ai off my post!
300 points
@oliver_io_ lazy* cheaters (eg slop)
300 points
@Shadow_Rebbe i think libraries basically did do this for like 2 centuries. tbd, I think the computer era kinda fricked them up a bit, we'll see if they adapt
300 points
@TylerAlterman that's still bad! we want them to be superhumanly "aligned" (eg. universally moral?), since humans are extremely "unaligned" (eg. adversarial), so if you give them a model like this, they will put pressure on it to break things and do harm.
300 points
@alyssaleann Fwiw, this is still the fastest adopted technology in human history, so this isn't a very big problem for its adoption
300 points
@myhandle i don't know... this doesn't seem right either, you've made me so interested in WHY "we're excited to announce" is so common, though... For instance, plenty of understated but genuinely 10/10 startups will write a fundraising or product post like "we are excited to announce..."
300 points
@relic_radiation It wasn't around when I was in undergrad, which was much more about "correctness". Now it's more about beauty and parsimoniousness
300 points
@abossy @0x1F9ED unfortunately, something like that appears to be going on
300 points
(i probably messed up my context window somehow)
300 points
@IaimforGOAT yea, i wonder if there's an even simpler answer that doesn't cause any delay
300 points
@disconcision ah, brilliant, tyfys
300 points
@sporadica @ElectricBoo4888 @JeffBezos so the $100M gets given to scientists, researchers, companies, employees, and contractors... those people are welcome to spend it on goods and services (hotels, flights, equipment, etc...) that economic activity has ripple effects, and so on...
300 points
@chilledgnillets @TylerAlterman 'course
300 points
@liamdanielduffy This is a great format. You should do this for future classes, too.
for my 18th birthday my girlfriend (who was a few months younger than me) invited me over to her place "while her parents were gone" 😏 when I arrived she left the room and my best friend walked in dressed as Chris Hansen and said "What did you think was going to happen here?" https://t.co/RJRSFSmAGe
I never realized that drones would bring us into the disco warfare era https://t.co/6c5b3Z0PgD
the secret to flirting is that if she likes you you can basically say whatever the fuck and if she doesn't even the smoothest line won't do shit
i used to be a delivery driver. today my sister tells me that she was talking to her friend's brother and he said that SIX YEARS AGO a pretty girl delivered him pizza and he still hasn't stopped thinking about it. she showed him a pic and it was me
In NYC, IMAX 70mm Odyssey tickets are so in-demand that they're basically just running the theater 24/7 now https://t.co/Oc1kgZJU1t
omg im so embarrassed,,, the abyss was actually gazing at its friend right behind me 😬😬😭
@chaotictransfem I am confident I could engineer a measurably better life along every quantitative axis and still be left longing for what could have been
binary search https://t.co/LUYsO1eZh4
This is such bitter loser behavior holy fucking shit, you gotta be completely cooked to think this is a flex https://t.co/eXruYXiify
every chair is a cuck chair when you're not having sex
BESTIE: ooooh third date tonight, you know what that means ME: (nodding) failing the penultimate layer of the survivability onion BESTIE: ME: fourth date he kills me https://t.co/ADPPtFiwVl
psa to sensitive young men: as long as you're respectful about it, this is how most women feel when you ask them out even if they say no https://t.co/3KwuulqQ1j
I put LotR on in Italian to see if Gandalf refers to the Balrog with the formal or informal "you" in YOU SHALL NOT PASS it's the informal, which raises a second question: is it because he doesn't respect the Balrog or because they boyz (both maiar)
*taps the sign* https://t.co/0KrgEuRyrx https://t.co/TuVacHu74P
hot girls are necessary part of a thriving city ecosystem. hot girls inspire ambitions and call you to transcend. Hot girls make the world go round. A city without hot girls is like a sky without stars https://t.co/sXPUfVmQrN
every time https://t.co/ONvY9LxIHu https://t.co/Dj2OrP7hAi
“why can’t we have this” and the images are of jared leto’s villain lair from blade runner 2049 which conveys to the audience that he is a power-hungry narcissist who builds replicants because he craves having total control over an infinite source of disposable bodies with no rights whose labor he can extract in perpetuity for his own enrichment y’know what actually, you’re right, every office in sf should look like this
in grad school i used to compulsively download interesting-sounding textbooks from libgen and sort them by subject in calibre. read almost none of them. it flattens the incentive landscape in a weird way to have invisible digital access to everything. very unnatural. the human animal yearns to come across a book organically by combing through the shelves of a library or bookstore and being struck by the physicality of its cover like our ancestors foraging through the bush for delicious treats
when i was at cambridge a professor was teaching algebraic geometry and offhandedly made some claim that was not clear to me i asked "is that obvious?" he says "yes" i asked "is it obvious that it's obvious?" he pauses for a few seconds and goes "...no" https://t.co/b5eV4r6NEO
it's common advice to dumb down your vocabulary to reach a wider audience. I wanna know who's elevating it to cut more people out. who's using words that were only written down once in the 1400s and once in the dictionary. who's writing at a 75th-grade reading level
the mark of a good streaming service is an iconic show/movie that you can't get anywhere else netflix: kpop demon hunters hbo: succession united states library of congress: Silent Footage of Liza Minnelli's Third Birthday Party as Filmed by Famed American Lyricist Ira Gershwin https://t.co/ARhr9FLymY
if you are autistic, which famously involves deficits in understanding social interactions, i do not recommend developing an elaborate theology of social relations in which autistic people are Good and neurotypical people are Bad. i recommend developing psychic powers instead
@itsyaboidougie one of the funniest things I've been a part of and I had nothing to do with it
https://t.co/EvE1DxAveH https://t.co/1tjPxawcKA
@NathanImago gdi you're right. they shouldn't have used nerf guns
this is a very sobering read. main thesis is in section 5: > Our conjecture is that a modern society makes the third child expensive and childlessness cheap, even when GDP per capita is only a fraction of the frontier (i.e., the U.S. level). Children are expensive because a child must be equipped to function in a world of formal knowledge and credentials, which takes years of costly investment (Restuccia and Vandenbroucke, 2013, Doepke and Zilibotti, 2019). At the same time, children are no longer productive in the household, not even for smaller chores, having been replaced by “engines of liberation” such as the electric appliances now available even to poor households (this idea extends the intuition in Greenwood et al., 2005). Modernity also imposes tight career schedules, set by large-scale organizations, that collide with the biological clock and with the kin networks that used to share the work of raising children (Adda et al., 2017, Compton and Pollak, 2014). And modernity makes childlessness cheap, because an impersonal society imposes next to no penalty on those who have no children: pensions and health care do the work that sons and daughters once did (Boldrin et al., 2015), and lineage counts for little. > Goldin (2026) pushes this line further with an argument about a growing mismatch between genders. Modernity also changes gender roles unevenly. Women’s opportunities expand rapidly through education, work, and greater independence, while expectations about housework and childcare often change more slowly (Goldin and Katz, 2002, Goldin, 2021). This is particularly true as economies shift toward services, where women’s comparative advantages are better rewarded (Ngai and Petrongolo, 2017). The mismatch operates at both the extensive and the 18 intensive margins: more women remain childless, and mothers have fewer children on average. > This argument is attractive because it explains why East Asia and now Latin America have such low TFRs. These are the societies where the gender mismatch is largest.19 It also suggests an intriguing possibility: societies with highly unbalanced gender norms start with very high TFRs, as men impose their preferences (Ashraf et al., 2014), and end with very low ones, as women resist (Doepke and Kindermann, 2019, Doepke and Tertilt, 2018). A paradoxical reversal of demographic fortunes. ... > In short, modernity in its current form seems largely incompatible with replacement-level fertility.
This is sincerely the most romantic story I’ve ever read. I haven’t googled it. I hope it’s true, I hope it’s real, I hope they live long and happy together https://t.co/zQIyxxOLYM
bro. a beer. a female beer
It’s gonna be real funny when the first Chinese models start triggering the Claude watermark detector
Imagine having the moment you unlocked *thinking* on camera: https://t.co/C6FhDNP8aK
@Flydie_Lwood oh, he'd been practicing
I have started putting bounties out for evidence when people try to spread lies about Pangram on Reddit https://t.co/m5WLAxp8vJ
I can't date someone with a creative passion unless I legitimately like/respect what they make. I was insane to ever think it could be otherwise
we should bring back cutely ironic punishments that echo the nature of the wrongdoer's crime for example, whoever replaced the twitter photo grid with the carousel could be drawn and quartered
“I don’t like you, and here’s your permit” is the essence of a pluralistic, small-l liberal society. https://t.co/f9B7Y8dCd7
fucking her for company secrets. call that inside her trading https://t.co/TS5aJtLDZI
thank you @NYC_DOT for fixing this! concrete barriers installed, problem solved ✅ https://t.co/RpGgpFNXqH https://t.co/ZXTntNOiID
He just whiffed that thing outta the cup so swiftlt with his hands in his pockets. Elite display of a degen skill. While wearin slacks, at that. This is the most attracted I've ever been to Tom Holland. https://t.co/lS6Uw628Pk
my current thing is doing chores lovingly. keeping love in my heart. remembering that the task in front of me is an act of love and for who. romanticizing domesticity, they call it
@pissvortex duvall-like
it blows my mind how my friends who parent enter a different timeline... it happens very quickly. they have 1 kid, then the kid is 2 years old, then they have another kid, and all of a sudden they have a whole family and household and they’re moving through a different set of moves while people who don’t have kids are still getting drinks
if you believe this you are going to own yourself for no reason. that is not what people mean by "oversharing." let me say it in the most autistic way possible: oversharing is when you have not established common knowledge that the other person wants to participate in the computation you are attempting to enroll them into participating in. it is a failure to understand basic human protocols for establishing trust, rapport, and common ground. the message you are communicating when you overshare is "i do not understand how relationships develop gradually over time as we mutually discern how much and in what ways we can and want to trust and rely on each other"
Oh, this breaks my heart. This woman needs to be in a facility. Which is what her lawyers are trying to enable. Meanwhile her doctors want her to go to jail so she doesn’t successfully sue them for irresponsibly over-prescribing while she went deeper into psychosis You probably don’t hate the medical industry enough. Like yeah, modern medicine is a miracle and a marvel, but from the bottom of my heart, fuck every physician whose concern for their patients is barely-disguised-contempt that leads to listening too little & prescribing too much
we’re all going to learn that the way someone treats AI predicts the way they treat anyone they have power over, btw https://t.co/KLtob5TO8k
walking by a math department and lobbing a golden apple that says: ✨"Round to the nearest integer: 0.49999999..."✨
@tyyler47 crazy and im sorry but also i read this as “cumming all over me google” at first and i was totally baffled
Whenever I read the term "looksmaxxing" (m2m transexualism) I recall this classic 4chan comment https://t.co/uXunbkbRzd
@mishapathy but on the flip side: if she already likes you, you basically cant screw it up!
https://t.co/8a6RmKKJPf
@marshmarigold i agree because i think if he really was trying to cheat on her, why would he even tell her about it
A variant of this: human-verified writing started 5000 years ago and ended in 2022 and NO ONE talks about this https://t.co/YV3RwA3ysd
You get one pangram release before everything from this model gets classified as AI https://t.co/2rQgbD3iyj
it turns out lots of very clever people tried thinking from first principles already, in detail, for thousands of years, and that’s how they came up with second, third, etc. principles, which correct the failures of first principles https://t.co/7f2FjlsFep
it jut keeps happening https://t.co/7LNyYNHmhD
Gorgeous girl wearing a very short yellow dress & kitten heels walking up the subway stairs and her impeccably dressed boyfriend walks carefully behind her as a shield for her ass, which is so close to out
In my view we have a few different tiers of verifiability 1) programatically verifiable (near-free) - games, coding, math, cybersecurity, chip design 2) real-world verifiable (cost-or time bounded) - sciences: biology, chemistry, physics -physical world: material science, energy, aerospace, robotics, agriculture, pharma -forecasting: trading, weather 3) verifiable with human preference - writing, design, comedy, charisma, persuasion I expect most low-hanging fruit in (1) to be solved very quickly. Not sure how much longer before more solved math conjectures are simply uninteresting. I’m least certain about chip design being in this category, perhaps we hit a ceiling and require physics or materials breakthroughs to continue progress. My guess is we will quickly run into the limits of how well we can simulate each domain in (2). The time-bounded nature of real world verification may be the reason we don’t hit fast takeoff. Sim2real remains an elusive problem to solve when real-world data is limited. Part of the reason I don’t expect to live multiple hundreds of years is simply that I expect pharmaceutical progress to be time-bounded by the physical world. I expect the items in (3) to never really get solved to a superhuman degree, as success relies on an ever-shifting plane of cultural preference. People adapted to “good” AI writing and became annoyed at new stylistic tics that, in a vacuum, are not necessarily bad.
grateful i grew up before i could access an internet filled with vicious gender / dating discourse, back in my day you just watched romcoms and listened to love songs on your ipod and you were like "hell yeah being in love kicks ass i hope i get to fall in love someday"
my contribution is that it's very weird that the sun and the moon are the same apparent size in the sky, the consensus appears to be that this is just a coincidence but that strikes me as wildly unsatisfying. seems to me there ought to be an anthropic explanation and my schizo theory is that eclipses are somehow necessary for the creation of civilization, like they provide some kind of crucial stimulus to insight or creativity or something. here's sol riffing off of this
confusing that people interpreted this as a status hierarchy. part of my motivation for going into math was that every other field seemed too *hard*. math was the only field where i could make sense of everything in my head. always resonated with this von neumann quote: https://t.co/Er8AY11mVZ https://t.co/onBl9ac62A
was at this hippie event in boulder once, one of the top 5 most beautiful women i've ever met in my life told me she'd sped up the healing of a broken bone months faster than her doctors believed possible by meditating on it every day https://t.co/9c994OkhoC