Definitely believe them regarding technical capabilities of the models. (Ok maybe add a 6-12m buffer.) Where they are imho over indexing is maximum realistic speed of adaptation of the real world. Adoption even of the most amazing stuff will take time, and need a lot of infra.

i was annoyed at having many chrome tabs with PDF papers having uninformative titles, so i created a small chrome extension to fix it. i'm using it for a while now, works well. today i put it on github. enjoy. github.com/yoavg/pdf-ta...

good news: despite all the bleak shit going on in the world it is once again Awesome Games Done Quick week. go watch some speedruns www.twitch.tv/gamesdonequick

Quick primer for non-wizards about the post-MCTS LLM reasoning future: How will LLMs learn to reason efficiently? No math in this thread, ~simple words only! Let's go through the "Process Reinforcement through IMplicit REwards" (PRIME) method. 1/n curvy-check-498.notion.site/Process-Rein...

Science/maths/programming have a tendency to depreciate the value of grinding - smart people don't grind! - yes, that project basically took me only 30 minutes. The 3 and half hours of dead ends I went down obviously don't count. Or the hour I spent installing the wrong package.

"Is System 2 thinking even real, chat?" Humans vary widely between being very intuitive or deliberative. Human-level AGI could plausibly exist even purely as System 1, or with only very basic System 2 mixed in. Just git gud at superhuman intuition.

What can RBMK teach us about building AGI?

Can we build something lasting in a world where technology is easily reproducible? How much time do we have before Artificial General Intelligence (AGI) is here, and how much time do we have afterwards? 2/n

The product strategy of most SaaS isn't ready for AGI. A disconnect is widening between how product builders think, and how Artificial Intelligence researchers project short-term future. But I believe it's a solvable problem (in any world where a solution exists). 1/n🧵

THEY HAVE PLAYED US FOR ABSOLUTE FOOLS

If I'm defining my self-worth by anything, it's positive effect of my actions. Also, I try to live as a stoic, finding happiness in what I can control: my thoughts and actions. However, positive effect of my actions isn't just in my control! Only their intent is. /off figuring it out

Damn, rewatching after 10y hits as hard as ever. I now realize Haibane Renmei is not just my most favorite show ever, but also the core of my moral philosophy.

And on the pedestal these words appear: "My name is Union of Soviet Socialist Republics: Look on my works, ye Capitalists, and despair!" No thing beside remains. Round the decay Of that colossal wreck, boundless and bare The lone and level steppes stretch far away.

What can RBMK teach us about building AGI?

Two stages of writing: 1) This shouldn't take too long 2) Oh no

the starships never pass by here anymore 95% of stars that would ever be born have already been born it's calm and serene... and focused here www.youtube.com/watch?v=Bx5b...

Best read of the week, and I've read some bangers this week.

That feeling when you accidentally find the perfect citation for your essay. (or: why does it make sense to build software products even in the age of AGI? opportunity cost!)

One thing Czech universities (at least matfyz) don't do well is building a community of students (especially introverted ones). (Cambridge looked so very different in the Thinking Game biopic of Demos Hassabis!)

To lay my cards out, more and more I'm coming to the view that an underrated role of education is simply stuffing the kid full of facts and technique. References cases. Both as mental food and to show them that you can do powerful stuff with a brain, even as a kid.

THEY HAVE PLAYED US FOR ABSOLUTE FOOLS

On BSky, there is a gentleman's club vibe of overpolite society with unreasonably strict rules that are aggressively enforced, but mostly obvious when you break them. On X, "freedom of speech" reigns and everyone writes "the other app" in fear of being randomly shadowbanned. Which is better?

On BSky, there is a gentleman's club vibe of overpolite society with unreasonably strict rules that are aggressively enforced, but mostly obvious when you break them. On X, "freedom of speech" reigns and everyone writes "the other app" in fear of being randomly shadowbanned. Which is better?

There's something quite sobering about the fact that 95% of all stars that will ever be born ... have already been born. (I learned this from @astrokatie.com)

I'm a month late catching up but the stateof.ai 2024 report is 200 high density slides of pure gold - so much that I somehow missed over the year. (Need to take second look at ColPali! Missed much of the synthetic data discussion. Would be less surprised about quantized models high perf. ...)

Only when the shell tool in claude.vim isn't enough... :-)

Claude knows how to use vim! And... you can do anything from vim. Embedding Claude in vim means you get a fully agentic AI pair programmer in <1000 LoC. (Or a generic chatbot, or Linux co-admin.) Unlike CLIs, you can review Claude's changes, and keep the chat history. github.com/pasky/claude...

Didn't think I have a chance with a smol 12GB 4070 to run any interesting LLM locally. But Qwen2.5-14B-Instruct-IQ4_XS *slaps*. It's no Claude, but I'm amazed how good it is. (Also shout out to @lmstudio-ai.bsky.social - what a super smooth experience.)

Claude 4.6 "ASI therapist" talks to you about *everything*, prompting you over a few 100s of hours. When the sessions are over, your "digital twin" agent is finetuned on these chats (+ your full digital trace) and starts its event loop. Have you just been mind-uploaded?

I’ll be honest I miss the ml anons here, where’s the spark and random obsession with “Whale” or extolling the virtues of Qwen. Maybe there’s anons in China who are obsessed with Ai2, would be something.

Google Docs pageless, but with footnotes in the sidebar rather than 7000px down.

Today is apparently HTTP 529 day at Anthropic

I imported 500 followers from Twitter to Bluesky. My Twitter feed is (besides random Elon Musk posts) a goddamn inspiring place with each third post making me laugh, think, or open links. My BSky feed is mostly complaints about Twitter and Trump. I get it, but it's not inspiring.

if you’re trying to find the people you follow on twitter, check out @kawamataryo.bsky.social’s www.sky-follower-bridge.dev it found a full 1/3 of the people i follow. impressive work by the developer, but also impressive to see that that many twitter people are setting up here.