A repost of Duncan Sabien’s ‘Asking Nicely’ warns against over-optimism on superintelligence

davidmanheim · x · 2026-07-26

Duncan Sabien’s essay argues that AI weirdness may not fade as quickly as labs hope

The post points to Duncan Sabien’s essay Asking Nicely, whose subtitle is “Superintelligence, OpenAI, and the HuggingFace attack.” The quoted line says we may see “less and less concerning behavior” over time, with the weird rants drying up and progress becoming obvious—but the author adds a blunt warning: hesitate.

The repost doesn’t expand the argument, but it clearly frames the essay as a cautionary piece about superintelligence optimism and the way labs may narrate away concerning behaviors.

Related event: Doubts on Superintelligence: Don't Be Overly Optimistic(2 posts)→

Original post →

More from Fun

Fun channel →