Hinton calls building uncontrollable AI "very stupid"; Altman says he's confident it's safe

DavidSKrueger · x · 2026-09-23

Two opposite verdicts in the same week: Geoffrey Hinton, whose work made all of this possible, says it's "very stupid" to build AI beyond our ability to control it, while Sam Altman, who sells it, says he's "confident it can be done safely." The poster highlights the incentive gap — Hinton quit Google specifically so he could speak out, while Altman only makes money if you believe him. When the inventor and the salesman disagree, follow the incentives.

Original post →

More from AGI Musings

AGI Musings channel →