Meta researcher: AI progress is like Moore's law — long run, but it ends
rbhar90 · x · 2026-09-22
Meta researcher rbhar90 argues AI progress resembles Moore's law: massive capital influx lets models keep finding new axes of improvement, possibly for years or decades — but like Moore's law, it must eventually end. Moore's law transformed society without producing a singularity; he expects the same: math advances will yield many downstream gains, but everyday-life shifts will take time, especially outside tech circles.
More from AGI Musings
- Physicist Sabine Hossenfelder: AI will never discover new physics, its ceiling is a coding grad student — alexbilz · 2026-09-22
- Longman defines 230K words with just 2,000 — so a 7B model may be enough — burkov · 2026-09-22
- Hinton: US and China won't agree on election deepfakes — every government wants the option — victor_explore · 2026-09-22
- Literary critics have a duty to spot AI writing and deny it prizes, argues critic — birchlse · 2026-09-22
- Hinton says AI understands and has emotions; researchers push back — AlexTensor · 2026-09-22
- When everyone runs long-lived agents, where will they talk and how will they know who's who? — repligate · 2026-09-22