On taste

Coding agents somehow learned to build good-looking websites, although nothing checks whether a design is beautiful. What video editing can borrow from that: preference models, critics and loops.

GPT-4o's image generation turned the internet into Studio Ghibli for a week. When every style is one prompt away, the scarce thing is judgment.

AlphaProof and AlphaGeometry 2 together solved four of six problems from this year's Math Olympiad, reaching silver-medal level. What makes a checkable answer so useful for training, and what happens when there isn't one.

Google's NIMA predicts how people would rate a photo, as a distribution, not a single score. Modeling disagreement turns out to be the useful part.