Where the AI market is talking

Karpathy’s Pelican

Don’t worry. I’m sure they’re not training it to be good at things like writing database backends, financial services, logistics systems, user interfaces, or anything of economic value. As long as you’re not working on three.js specifically, I’m sure Anthropic isn’t making any progress you should be worried about.

The OpenAI and Anthropic AI Hacking Sprees Are a Messy New Legal Frontier

Nanocodex: Building blocks for frontier OpenAI agents in Rust

The Coldcard Disaster Gets Worse: The Hack May Have Reached $88.6M

That’s because it’s, at least partially, authored using an LLM. From experience it reads like Anthropic’s

Karpathy’s Pelican

Karpathy works for Anthropic so he obviously has full access to everything, and doesn’t have to pay for token burn either.

Karpathy’s Pelican

Anthropic spokesman [0] Andrej Karpathy is here to tell you about token-wasting loops, and insists on the weird idea that they are "~free", when in fact, they are fuelled by expensively burning investor money. [0] Seriously. Get used to mentally prefixing his and Boris Cherny's name like this, every time you see them quoted. These people are speaking while employed; there is no chance they are not aligned with the employers who will make them wealthy. The tech industry does like to pretend that for some reason AI people, uniquely, speak thoughts unbiased and for themselves or even for science or humanity.

Karpathy’s Pelican

I’ll never not be amazed that we can type some words and get those results back out. However I do agree that the results are much worse when you try to use them. They look great in screenshots and video clips which makes them perfect for content farmers. All of the LLM generated games I’ve played have been really bad to play, though. I even tried my hand at a simple game, thinking I could iterate on it with prompts to fix some rough edges. After the initial productivity burst every change turned into a slog of tokens with one thing changing and something else breaking it. I would try to use my remaining weekly token budget across Anthropic and OpenAI to refine it at the end of every week but after a couple weeks it felt like I wouldn’t be getting anywhere without scrapping it and going ba…

Karpathy’s Pelican

I’ll never not be amazed that we can type some words and get those results back out. However I do agree that the results are much worse when you try to use them. They look great in screenshots and video clips which makes them perfect for content farmers. All of the LLM generated games I’ve played have been really bad to play, though. I even tried my hand at a simple game, thinking I could iterate on it with prompts to fix some rough edges. After the initial productivity burst every change turned into a slog of tokens with one thing changing and something else breaking it. I would try to use my remaining weekly token budget across Anthropic and OpenAI to refine it at the end of every week but after a couple weeks it felt like I wouldn’t be getting anywhere without scrapping it and going ba…

Ten advances in mathematics and theoretical computer science

It's disjointed? The post that started this sub-thread asked: > 1. How many total problems were given to the model, and what percent were left unsolved at what cost before giving up? 2. How many attempts did you give the model at solving these problems? 3. How expensive was the harness, e.g. did the model have access to a job cluster? I think it's an extremely relevant question to ask, because it helps us better understand the current state of AI being able to handle math, for exactly the reasons I outlined. I was arguing against the idea this is just a reactionary anti-AI kind of question to ask. It's not! You can be very impressed by what AI is capable of in math (I am) and still think those are really interesting things for OpenAI to disclose (I do). OpenAI specifically called out a…

Rust All Hands 2026 Retrospective

Just run Rust compiles in the cloud. Really, if you're using a SaaSS LLM like Claude or Copilot or ChatGPT, which already sends all your code to someone else's computer to run on a beefy GPU, they should just send the Rust code to the same datacenter and send you back a binary. Little shocked that Anthropic isn't offering this. Rust compile speeds will matter less and less as hardware gets faster.

Karpathy’s Pelican

It seems pretty clear that Anthropic models have been specifically trained to be good at generating three.js (JavaScript 3-D Graphics) code, so given current state of AI code generation in general, I don't find three.js models/animations as indicative of anything other than the model's ability to write three.js code. When Fable was first released the day-1 demos of it on Twitter (presumably from people who were given early access, and/or Anthropic employees) were pretty much 100% three.js stuff. Yes, it looks nice, but it doesn't tell me any better than an Erdos proof whether the LLM will be able to run my vending machine.

Karpathy’s Pelican

I'm sad that Andrej Karpathy went from being one of the most reasonable, trusted, and credible voices in AI to peddling marketing slop for Anthropic. 8 months ago, he was (very reasonably) claiming that reliable agents are at least a decade away, but this now goes against the interest of his employer, so the narrative has been changed.

Running Kimi K3 on MI355X at Better Performance per Dollar Than B300

hi I work at wafer. yes we ran benchmarks. for example our Kimi K3 is live on open router and in order to host there you have to run accuracy checks. like tau/gpqa. and then u must pass test regarding thinking, coherency, and tool calling

Anthropic's Opus 5 Is Better at Resisting Prompt Injection

An internal OpenAI Astra model solved 10 major open math and CS problems

Go back and re-read. Did I say anything bad about AI? What I'm talking about is the way OpenAI have handled this. They are in the business of artificial intelligence, striving to create human-level, or even super-human intelligence, and telling us what a wonderful future this will create. Shouldn't they be celebrating intelligence, especially human intelligence which is what they have bottled here?! LLMs are themselves minimally intelligent - they are primarily regurgitating human intelligence. Without their human training data, these systems would be nothing. Let's see OpenAI build "GPTZero", an actually intelligent system, that learns for itself, learns language and mathematics from nothing like a baby, then I feel they'd have the right to minimize and cheapen the achievement if they ch…

Zitron: "Everyone Has Been Sold a Lie" on AI [video]

Good point. We can always point to Naomi Wu as an real life example of why China is NOT living in the future. She is a good example of why we would never want to replicate what China is doing

The session you cannot take with you

This path would only be viable for one or two providers and even does not work within one provider. For instance within OpenAI many of their models do not share the same capabilities. The complex of what you are suggesting is staggering considering the number of providers and models involved.

Go 1.27 Interactive Tour

On the contrary, linq in C# is a killer function that other languages still fail to replicate. How many sloc are required for this? var max = mycollection.Max(); Probably 5-10 if you’re missing higher order functions. And the risk of bugs will be 10x.

China begins producing advanced chipmaking deep-ultraviolet lithography machines

Okay, I think you’re right about some things. Coordinating a market economy is not straightforward. Individual participants will buy what is optimal for them. By embargoing them we provided a coordination mechanism. Chinese semiconductor manufacturers now had a captive market to sell into and the state was able to incentivize production without too much waste. But the thing that strikes me is that China was already trying to do this. SMIC et al were already funded by the Big Fund. And they generally succeed in industrial policy. It’s true that a GPU can’t be torn apart and replicated like a car factory can. But they were already trying and they were already attracting sufficient numbers. I suspect we couldn’t have stopped them. DUV tech like this has been already explored deeply. And the …

The Vanishing Page: AI Firms Scan Then Destroy Rare Book Editions

When are you or anyone more likely to get access to a rare book: a) When a company has made a digitized copy of it b) When a few previous hard copies are kept somewhere in the world Idk what Anthropic plans to do with this, but note that my original comment was not about Anthropic at all. I simply think it's a good tradeoff to destroy a few copies of rare books if that helps digitizing them. I am not against simply legally forcing companies that do this to release those copies in due time. In the meantime I am also happy to have them as part of the LLM corpus. > illegally? No, copying rare books is very likely not illegal, because, presumably, they are quite old.