On Cory Doctorow's new post: real agent risks vs. disembodied p(doom) framings

mmitchell_ai · x · 2026-09-16

Cory Doctorow published a new post on whether AI will kill humanity. Commenter agstrait argues both that AI agents are far more complex than a Python loop and carry serious risks of unreliable, uncontrolled behavior at scale, and that bullshit p(doom) framings disembodied from the tech companies causing these risks deserve critique. Safety researcher Margaret Mitchell amplified the take.

Related event: Debate Reignites Over Whether AI Could Destroy Humanity(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →