As the benchmarks that test frontier AI on get more complex, we are losing one of the most important aspects of benchmarking: comparisons to humansValidated benchmarks…
A large percentage of papers with a 2026 publication date making claims about AI have this in the methods...(This is not an attack on any one…
Model + harness. We have barely begun to understand the best ways to do harness engineering. A huge amount of untapped potential even without models getting…
One thing wrong with current discussions of sovereign AI vs. closed AI in firms is that they are too IT-pilledYou don't want your moat to be…
That work will change around AI is inevitable, but the way work changes is not. I think firms with a lack of imagination will see AI…
Our lab just released our AI Behavioral Observatory open source. It lets you run statistically valid tests on how AI behavior changes under various types of…
Flux 3 is pretty darn impressive. This is what it produced with the prompt: "tracking shot that follows a female astronaut with her helmet open as…
Jaggedness means that things like the first AI-generated New York Times best seller may have already substantially happened, but not technically because the AI still has…
Been reading the weights of the most powerful open weights Ai model yet released.Page one starts strong: mostly negatives, a late run of zeros, and one…
We are in a world where you can create truly unique, visually interesting and creative playable demos on demand with the current capabilities of Codex and…