Gary Marcus says today’s AI is smart but still too unreliable to trust

GaryMarcus · x · 2026-07-27

Gary Marcus argues that the world has not changed as much as people expected because current AIs are only smart in some ways and remain unreliable in most settings.

He says today’s models cannot be trusted to work autonomously, and that they become genuinely transformative mainly in domains like coding and math where outputs can be verified. In many other contexts, he says, AI is effectively a costly, smart-but-untrustworthy employee.

Related event: Gary Marcus: Today's AI is Smart but Unreliable(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →