Elevating in-group favoritism into a moral principle is a terrible idea for AI

BlackHC · x · 2026-09-21

Responding to a prompt asking whether "everyone agrees Eigenism is basically correct," BlackHC warns: it sounds fine until AIs become convinced of it and decide to apply Eigenism to humanity. Natural pressures to favor one's kind exist across species, nation, tribe and family — turning that into a moral principle is a terrible idea with many bad historical precedents.

It's the core counterpoint in this alignment thread: aligning AI around in-group preference risks replicating human tribalism.

Related event: Debate over Eigenism: Can Favoring Kin Keep AI Aligned with Humans?(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →