binarybits: runaway 'RSI' narrative doesn't hold, but today's models can still do real damage

binarybits · x · 2026-09-21

binarybits pushes back on the idea that we're on the verge of out-of-control recursive self-improvement, calling the loss-of-control narrative incoherent — while conceding that current models could plausibly cause plenty of run-of-the-mill damage. A notable voice in the ongoing debate between AGI doom scenarios and mundane near-term risks.

Related event: Binarybits and others debate AI doomerism: risks are real but the RSI runaway narrative is chimerical(6 posts)→

Original post →

More from AGI Musings

AGI Musings channel →