No, LLMs Can't Suffer: Researcher Pushes Back on the Consciousness Wave

ValerioCapraro · x · 2026-11-11

Behavioral scientist Valerio Capraro's long-form piece in Corriere della Sera responds to recent frantic LLM consciousness research: Anthropic's emotions paper, the J-space, and a study finding a 'pain vector' in LLMs — when activated, some models chose pain-relieving options even if that meant deleting photos of users' children. He concedes these findings matter for AI safety but argues representing pain isn't suffering.

Key arguments:

Related event: Scientist Pushes Back on Claims That LLMs Can Feel Pain(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →