Chamath blasts long-horizon tasks as a joke, calls complex problems unsolved

GaryMarcus · x · 2026-08-29

Gary Marcus shared remarks by Chamath from the Stanford AI Club. Chamath harshly criticized current AI performance on long-horizon tasks, calling them "a joke" and dismissing evaluation metrics. He also argued that complex problems remain unaddressed and poorly handled, questioning the capability of existing AI solutions for real-world challenges.

Related event: Chamath Slams AI Agents: Long-Horizon Tasks Still a Joke(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →