Understanding AI dissects Yudkowsky's new book: the case for AI doom isn't convincing

binarybits · x · 2026-09-22

Timothy B. Lee reviews Eliezer Yudkowsky and Nate Soares's new book If Anyone Builds It, Everyone Dies at Understanding AI. He notes the striking fact that insiders like Dario Amodei ("25% chance things go really badly") and Sam Altman take existential risk seriously, then lays out the doom argument's three steps: superhuman AI is coming, it will gain real-world power, and we can't ensure it's used for good. Lee finds skeptics of step one too complacent, but disputes the later steps, and warns that lab insiders' opinions suffer from motivated reasoning.

Related event: Understanding AI critiques Yudkowsky's AI doom argument(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →