'Sutskever instantly saw superintelligence': researcher grapples with failed frontier expectations

teortaxesTex · x · 2026-09-20

Researcher teortaxesTex expresses disbelief over the apparent disappointment of the Astra/Fable project, noting that most people at frontier labs expected it to obviously succeed and that Ilya Sutskever himself 'instantly saw superintelligence' in it. He argues that if Sutskever was wrong, the epistemology behind expert intuition is worth questioning, while criticizing skeptics who demand first-principles proof — an attitude he says suits safety engineering but is bad for understanding emergent historical transitions.

Related event: AI community debates short timelines and failed superintelligence predictions(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →