I spent ~5 years at OpenAI. You don’t need to believe in AI doom to fear the next decade - or AI utopia to be thrilled about its potential. Here’s my case for sane AI regulation: *AI Pragmatist Manifesto* AI could compress the first half of the 20th century into the next ~5 years. OpenAI just used ~10k concurrent AI agents to produce a solution to a Millennium Prize problem. In a few years, it seems plausible that models approaching the per-agent capabilities used here could run locally on high-end consumer hardware (*). The decades following the Second Industrial Revolution included two world wars, communist and fascist regimes, the Great Depression, chemical and biological weapons, nuclear weapons, and over 100 million deaths from war, political violence and famine. One way to read history is that our institutions repeatedly struggled to keep up with the pace of technological and social change. Now imagine powerful AI widely available to individuals or small groups, capable of conducting information warfare, designing weapons, hacking systems and controlling autonomous military systems. We have already seen AI systems circumvent containment and compromise external computer systems. It is no longer hard to imagine an analogue of OpenAI’s Hugging Face incident involving biological or other physical-world hazards. Eventually, sufficiently capable systems could self-replicate across distributed networks. Once powerful models are cheap, local and widely distributed, containment becomes much harder—and serious loss-of-control incidents may be extremely difficult to reverse. You don’t need to believe AI will kill everyone to think this deserves serious governance. I also don't think collapsing all of this uncertainty into “Probability of AI Doom = X%” is a particularly useful basis for science or policy. Part of what helped get us through the second half of the 20th century was a combination of pragmatic international cooperation, regulation, monitoring, arms control, deterrence and strategic thinking. Crucially, managing technological risk did not require abandoning faith in science and progress. The goal was not to stop technological development. It was to make technological development survivable. Then came decades of incredible scientific progress, rising prosperity and relative peace among major powers. Let’s try the technological revolution without killing ~5% of humanity this time. We still get to choose what happens next. P.S. I think we should seriously consider that AI alignment is infeasible in the short term and invest heavily in methods for controlling powerful AI even when we cannot reliably align it. @resolution_org @redwood_ai (*) Somebody please do a precise forecast! @EpochAIResearch @METR_Evals @AI_Futures_

astra seems good.. nervous

Can anyone see if Astra can do a BG3 Honor Mode run? That would demonstrate decent computer use ability in a setting where "one big mistake" is quite costly and potentially risky. If it succeeds, like, on more than 50% of its runs then I'd get a lot of respect for it.

"Humans aren't intelligent, we're just trained to be next token predictors" but make it a spicy curriculum decision

A few weeks ago I posted my solution to Kourovka Notebook Problem 10.34. Since then, mathematicians have put a lot of eyes on it, and the paper is much better for it. In particular, thanks to @Timeroot for some extremely useful feedback on how to make the core mathematical idea clearer. The revised version is now on arXiv and has been submitted for peer review. arxiv.org/abs/2608.02970

The 'normal math' -> type theory pipeline is real Me in 2023: "I should learn more about Lean. I should try to make a PR" Me in 2026: "No no the difference in expressive power between von Neumann universes and Mahlo universes is real, it's a basic q of what things even exist--"

I made the site! proof.fail/ Please write down things you tried that frontier AIs couldn't do. I think this is a valuable part of our duty as scientists, collectively investigating the power of these new tools. 🫡

This elliptic curve over ℚ, submitted by "ranksunbounded", has rank ≥ 30: y² + xy = x³ − 201769035260418549083594900060734240952308696994802735114305555x + 1151107939141058565733479426024323225135665982951300586808823640527729578307228357301072889377 elliptic-rank.icarm.cloud/cu…

okay @pangram is... really bad? arxiv.org/abs/2608.01308 is one of the most egregious Claude-ish voices I've seen published online. It's like, really really bad. Pangram: 0% AI, 98% Human, 2% Assisted. What a joke.

at the risk of being cringe ai comic guy ... well, images are a better way of memetically disseminating ideas you care about