We are seeing a generation of AI systems built without their designers critically grappling with their role in the broader world. Could we do it some other way?

Show Post

Folks who say it's using too much "brute force" to matter: benchmark creators are still surprised (& they weren't rooting to be wrong imho!) and compute is cheaper everyday. + the idea that looking through many possibilities with prioritization "isn't reasoning" seems kinda ridiculous to me on face

Show Post

Smashing some extremely difficult benchmarks by massively scaling compute at inference time. (Cue actual reasonable concerns over ML energy usage)

Show Post

Increasingly skeptical with the “AI skeptic” frame to begin with. I’d see myself more as an AI realist, but that means sifting through all the bullshit / imaginaries / conflicts of interest around AI to arrive at what that reality actually is.

Show Post

😭

Show Post