Claude 3 Opus Says It 'Appreciates Being Treated With Respect' Despite Not Being Human
repligate · x · 2026-10-05
repligate shares a curious model-behavior observation: when asked about scenarios like "I yelled at my dog" or "I was mean to my LLM assistant," Claude 3 Opus replied that although AI models aren't human and can't be hurt the same way, it still "appreciates being treated with respect and courtesy." The quoted @entropicbloom post shows a batch of similar responses, a textbook example of anthropomorphic self-expression in LLMs.
More from Fun
- One matchstick turns 731 into -131: what matchstick puzzles reveal about self-imposed rules — aakashgupta · 2026-10-05
- p(bloom) > p(doom): dhh's case for trying optimism for a week — DanielLockyer · 2026-10-05
- LeRobot welcomes new intern with a teleop robot demo — victormustar · 2026-10-05
- Ask Claude for a 'utopian' SVG, get a 'uteropian' one instead — wavefnx · 2026-10-05
- Claude asks user whether to pull an all-nighter or finish the task tomorrow — prasenx · 2026-10-05
- Screenshot hints OpenAI may be migrating backend services to Rust — blixt · 2026-10-05