Reaction
The joke about AI timelines where everyone else is wrong
A post jokes that timelines shorter than the author's miss real-world bottlenecks, while longer ones ignore recent capability gains.
TLDR
A post pokes fun at treating one's own AI timeline as the reasonable one: shorter timelines supposedly ignore real-world bottlenecks and diffusion challenges, while longer ones supposedly make “basic Bayesian errors” by not updating on recent capability gains.
Combined views
613
2 Sources, first seen ago