Rereading the Hanson-Yudkowsky doom debate: the em assumption does all the work

herbiebradley · x · 2026-09-24

Researcher Herbie Bradley, having reread the Hanson-Yudkowsky AI doom debates, notes that a recent argument is essentially substituting ems for AI — but that substitution works best when the entities are perfect human emulations with human drives. We already know LLMs lack human drives and have a differently shaped intelligence, so the "human-like" framing is misleading while quietly doing all the argumentative work.

Original post →

More from AGI Musings

AGI Musings channel →