@ShadowCodr @dogmycatsjoe @cursor_ai I knew you'd be too much of a coward to even give an actual reply to my question. People like you are always just ignorant cowards.
English
Hal Wayland
77 posts
Hal Wayland
@Hal_Wayland
If I won a lottery, I would learn computational geometry, code it in Odin and Metal, and just have fun with it for a few years. I'm doing other stuff though...
Slovakia 参加日 Mayıs 2026
284 フォロー中4 フォロワー
We're launching code storage and git hosting.
Origin gives teams and agents a place to host, review, and collaborate on code.
Available this fall. Join the waitlist.
cursor.com/origin-waitlist
English
HE DID IT
THE FUCKING MADMAN ACTUALLY DID IT
Schwabster🇱🇹🇩🇰🇺🇦@Ziggydank
Only a German mind could cook up such a niche
English
English
English
@nonzeroexitcode Jarring and completely unnecessary, don't you think?
English
another super-hover demo!
gonna milk this for a while because it works well for lists, especially work-section for portfolios
and it’s just fun to play with:)
English
@morganlinton @mntruell All I need is for it to be as fast as the current Composer models. I hope things won't get too slow now.
English
@peduarte Are people really going back to CSS-in-JS? I thought that entire idea is dead already.
English
didn't know linear was already migrating to stylex too
Adam Hutchinson@adjohu
We've moved more than 50% of Linear to StyleX so far. It's given us more control over how components compose, what can be overridden, and improved render performance too.
English
@peduarte I have the same dilemma, the problem for me is that Composer is literally the best model for me, by far, simply because of how fast it is. Nothing really comes close in terms of ability for TPS for price. Kimi on Cerebras would do it but consumers don't have access to that.
English
bittersweet moment
60B is an unfathomable number
on one hand, i'm happy for my friends at cursor. i hope many will be able to retire in the next few years
on the other hand, as hypocritical as it seems, not sure i fancy using another elon product
English
@dogmycatsjoe @cursor_ai I can't believe some people are still not over this. Just use main and shut up about it.
English
You gotta love when things like this appear in your feed right next to each other @mitchellh @arena
English
Hal Wayland がリツイート
@GergelyOrosz If only the verified badge actually meant something, that would be cool.
English
Starting to get sick of anon accounts that take bits from random articles (probably w AI), put them together into a piece, and then post it. And it's somehow retweeted.
The below is all AI slop. Devs don't generate puzzles: they do RLHF. Not 6,500 of them etc. Junk.
Hedgie@HedgieMarkets
🦔Meta moved 6,500 engineers into a new Applied AI unit through surprise emails three months ago. Their job is to generate puzzles and coding problems to train Meta's AI models. Employees describe the work as "soul-crushing." One called it "the gulag." During a livestreamed internal presentation this week, someone hijacked the feed with an expletive-laden meltdown directed at a senior AI executive. Over 1,600 employees signed a petition against a program that monitors their clicks and keystrokes for AI training data. My Take Meta cut 2,212 people from Menlo Park a few weeks ago because the budget had to go toward AI. Meta drafted the surviving engineers into a unit that trains the AI. They know what the work produces. That's why they're angry. Zuckerberg said he picked internal engineers over contractors because they're smarter. Meta paid $14.3 billion for Alexandr Wang's data-labeling company Scale AI, put him in charge of Meta Superintelligence Labs, and then drafted its own engineers to do the data labeling anyway. The unit is led by a VP who ran Reality Labs, which burned $83 billion on the metaverse before Meta moved on. These are the people and the track record behind the AI bet Meta put $145 billion into this year. Hedgie🤗
English
This is what I want my robots to looks like - a machine
Polymarket@Polymarket
JUST IN: Genesis AI unveils “Eno,” its industrial humanoid robot with no head or legs.
English
AI, just like oil in the 20th century, is about to become the major source of leverage and power in the world. Depending on how the coming years unfold, it will either lead to a world of wealth and abundance for all, or to the worst extractive economies that the world has ever seen. We're there to fight for the first scenario, as we progress AI research and accelerate its diffusion across the world – we're hiring if you like the quest.
English
We somehow got put in the spotlight the last few days! First we'd like to thank the organizers of the AI show for that, we can't get enough of this stuff. I'll say a few things about where we are and what we do.
English
@stalman It's literally not your photo anymore, by definition. What's important is whether or not you actually care in case of that particular image.
English
I think I figured out what bother's me about Apple's new Reframe feature: have you heard of the Star Trek teleporter paradox? Or seen The Prestige?
Even a small adjustment means AI rebuilds a new image from scratch. It looks like your photo, but is it really your photo?
English
@denbvk @ThePrimeagen Seems like you need to find more creative and productive friends then. The friends you have now are clearly not capable to do anything unless it's playing with toys - what could be easier than giving a vague description of some project to an LLM and watch it produce something?
English
@Hal_Wayland @ThePrimeagen I’ve seen friends building some pretty useless pet projects with LLMs. Before, they did not build anything. I guess this is progress.
English
All the devs whining about fable being banned need to go hit the gym and stfu
English
Good question. I've seen a lot of genuinely impressive things being build over the years without the LLMs. Those people are still making impressive things and they are usually still not using the LLMs, even now.
On the other hand, I can't think of a single impressive useful project someone made with heavy use of LLMs. I might be forgetting something but I can't think of anything.
Personally, I actually do use them for work. I use DeepSeek V4 Flash or something that's similarly fast, I check all the output and confirm I would write that, the usual "use the LLM as a faster way to write code" kind of approach. I'm building internal tooling using web dev tools for EV charger support, just for context. Nothing "impressive" either though.
English
@Hal_Wayland @ThePrimeagen Have you seen lots of people producing anything useful without it? I mean outside of a bunch of famous projects. Like your friend suddenly building something truly amazing outside of their work?
English
@willchen500 If this is anywhere near true then Europe is truly screwed. Sounds like something that we would do as well.
English
I once talked to a Mistral researcher who told me that even internally everyone just uses Claude Code or Codex.
I asked him how does the company survive given that it's models are so far from the frontier and he told me that Mistral hovers up contracts with all the major EU enterprises like Airbus and BMW. Likely with French government backing.
It is remarkable that the Germans and the English are letting the French get away with this. It seems both countries are too caught up with their own domestic issues.
If there were actual competition like in the Chinese ecosystem I imagine EU AI sovereignty might be somewhat of a realistic prospect.
English
@thebriangao You've written the renderer code yourself on a piece of paper and had Fable read it or something? What the hell does "hand-written renderer" mean here?
English
I had Claude Fable 5 build Minecraft from scratch.
45k lines of Swift + Metal. Hand-written renderer, every sound synthesized, fully deterministic, zero dependencies, no game engine.
Open source 👇
github.com/thebriangao/pe…
English
@ThePrimeagen And I still haven't seen anyone really produce anything useful with any of it.
English
btw, this is exactly what I am talking about
"GTP5.5 is so bad, its trash"
No its not. You are such a loser.
Its the same dumb cycle every time. The moment something bets X% better (and from my experience it won ~4/5 prompt offs with gpt, so it felt better) the previous thing's value goes from "is this agi?" to "its trash" SOOO quickly.
English
@laurenloprete That's the most beautiful numpad I've ever seen, that can't even be real. Just look at how perfectly sized and spaces those numbers are.
English
People always ask me “how do you develop taste?” And I say “go to the niche telephone museum and study the origin of buttons and don’t be a little bitch”
English