Stanford's Omar Khattab coins 'virality law': humans and LLMs don't read, they average k-nearest patterns

lateinteraction · x · 2026-11-21

Stanford professor Omar Khattab observes that past a certain scale of reach, clarity stops mattering: people and language models alike don't really read — they seem to just average the k nearest patterns they've felt before. He jokingly calls it 'khattab's virality law,' a quip that doubles as commentary on both viral misreadings and LLM behavior.

Original post →

More from AGI Musings

AGI Musings channel →