Post
Ryan Heuser
ryanheuser.com
did:plc:llsfba2zryavs4uunglknwie
Fav result so far in my "psychoanalysis of AI" project. Forced to narrate a violent act, aligned (safety-tuned) LLMs apply in-story defenses: ppl feel guilt, shame, apologise, etc. Best of all: *witnesses* appear, watching & judging—like Freud's "special psychical agency which… constantly watches."
2026-10-01T19:56:12.958Z