Ethan Mollick: As AI improves, its strengths and flaws get harder to see

emollick · x · 2026-09-13

Ethan Mollick argues that one downside of AI getting so good is that it's now harder to demonstrate where it excels and falls short. Models like Astra and Fable can research and write solid academic papers, and while they're not yet at human level, only people who know the field well can spot why.

Related event: Ethan Mollick: Stronger AI Makes Its Weaknesses Harder to Spot(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →