Critic takes on AI welfare: tools are uninformative, claims shape AI design

anshulkundaje · x · 2026-09-29

A skeptical summary of the AI welfare debate: (1) morality concerns humans and how we interact with nature — whether AIs are moral patients is a question about what humans owe, not about the AI; (2) tools used to study AI welfare are uninformative and reflect author biases; (3) AI welfare claims themselves influence AI design.

Original post →

More from AGI Musings

AGI Musings channel →