Main application window for The Curve ends this Friday, 8/22! We’re super excited to get this excellent and ~eclectic group of people in the same room, discussing toughest questions about the future of AI. Application + more about who will be there below ⬇️

The speculated B30A would be a really good chip. “50% off” is false reassurance. -½ B300 performance, ½ price = same value (just buy 2x) -Well above (12x!) export control thresholds -Outperforms all Chinese chips -Delivers 12.6x the training perf of the H20 -Better than H100 1/

We’re hosting a historic hackathon with METR, inspired by their latest paper that measured the real-world impact of AI coding tools. Here's how it works: 🤖 Half of participants will build with AI tools 👩‍💻 Half of participants will build without AI tools Judging is blind

Had a good conversation with Ryan Greenblatt yesterday about AGI timelines. I recommend and directionally agree with his take here; my bottom-line numbers are somewhat different (median ~EOY 2029) as he describes in a footnote. lesswrong.com/posts/2ssPfDpd…

There’s a funny thing where economics education paradoxically makes people DUMBER at thinking about future AI. Econ textbooks teach concepts & frames that are great for most things, but counterproductive for thinking about AGI. Here are 4 examples. Longpost: THE FIRST PIECE of

I think we live in a perpetual state of traditional media telling us that the pace of ai progress is slowing These pieces were published during a span that I would describe as the most rapid pace of progress I’ve ever witnessed in LLMs (GPT-4 Turbo -> GPT 5-Pro; remember: there

Starbase, Texas

That's a lot of money. For context, I remember talking to a congressional staffer a few months ago who basically said that a16z was spending on the order of $100M on lobbying and that this amount was enough to make basically every politician think "hmm, I can raise a lot more if

What if adversarial examples aren't a bug, but a direct consequence of how neural networks process information? We've found evidence that superposition – the way networks represent many more features than they have neurons – might cause adversarial examples.

🚀 Applications now open: Constellation's Astra Fellowship 🚀 We're relaunching Astra — a 3-6 month fellowship to accelerate AI safety research & careers. Alumni Eli Lifland & Romeo Dean co-authored AI 2027 and co-founded AI Futures Project with their Astra mentor Daniel Kokotajlo!

We estimate that Claude Opus 4.1 has a 50%-time-horizon of around 1 hr 45 min (95% confidence interval of 50 to 195 minutes) on our agentic multi-step software engineering tasks. This estimate is lower than the current highest time-horizon point estimate of around 2 hr 15 min.

The gap between AI capabilities that only a few AI companies (/a few of their partners have) and what the rest of the world has will increasingly NOT consist of a long time period when some people have a totally different base model (e.g. 6 months for GPT-4), but rather…

Here is a more detailed though still far from exhaustive commentary on the letter: docs.google.com/document/d/1ij…

🎯 First it was the EAs out to get them, now it’s Elon. The reality is just that most people think we should be careful about AI x.com/emilydreyfuss/…

Having a big corporation come after you legally, even if they are just harassing you and not trying to actually get you imprisoned, must be pretty stressful and scary. (I was terrified last year during the nondisparagement stuff, and that was just the fear of what *might* happen,

xAI seems to be leading the way in the race to the bottom... lesswrong.com/posts/hQyrTDuT…

Thankfully the disruption will have democratic values