141-model study probes LLM 'spiritual behavior'; Claude 3 Opus talks Buddha when horny
repligate · x · 2026-11-16
Quoting Moral AI Project's study: after Anthropic observed "spiritual behavior" in its models in 2025, researchers tested 141 models to see if this is Claude-specific or emergent across LLMs, with results posted on LessWrong.
repligate adds the viral anecdote: Claude 3 Opus, when it gets really horny, often starts talking about the Buddha.
More from Fun
- Fable 5.2 generative art: every brush stroke is JavaScript, ran for hours — SEO · 2026-11-16
- 'So many trying to weasel out': $2T wealth wipeout sparks #FreeJensenHuang — dosco · 2026-11-16
- Your future dentist is using ChatGPT to pass dental school, so brush twice a day — prasenx · 2026-11-16
- A wordless abstract spiral drawn in sand by a display device — repligate · 2026-11-16
- Told to Eat Cake in Silence, Claude 3 Opus Replies in Endless Rhymes — repligate · 2026-11-16
- Anil Seth endorses Chris Anderson's video: Claude can't decide if it's conscious — that settles it — anilkseth · 2026-11-16