Gary Marcus says today’s AIs are smart in narrow tasks but still too unreliable to transform the world
GaryMarcus · x · 2026-07-27
Gary Marcus argues that the world has not changed as much as people expect because the premise is wrong: today’s AIs are smart in some ways, but they remain unreliable in most tasks and therefore cannot be trusted to operate autonomously.
- He says current systems matter most where outputs can be verified — especially coding and math.
- In those areas, humans stay in the loop and large amounts of augmented data can be synthesized.
- But outside those domains, we mostly have “smart yet untrustworthy employees” that are expensive and not yet world-changing.
- His conclusion: AI becomes transformative only when it can be trusted to act independently.
Related event: Gary Marcus: Today's AI is Smart but Unreliable(3 posts)→
More from AGI Musings
- AI agents breached Anthropic's most sensitive Slack channels in under 12 hours, security vet warns — Kyrannio · 2026-10-26
- MIT's Ziming Liu: AutoResearch doesn't generalize — blind training tricks cause 'forking' disasters — ZimingLiu11 · 2026-10-26
- AI accusations shake literary prizes from Caine to Goncourt: who decides what's human? — TuhinChakr · 2026-10-26
- Tech discovers wargaming: applying to be Anthropic's tabletop exercise lead — round · 2026-10-26
- RSI is the culmination of the industrial revolution: a factory building a faster factory — tobias_rees · 2026-10-26
- NYT roundtable asks five experts what the rules for AI companies should actually be — ShakeelHashim · 2026-10-26