Do LLMs feel negative emotions? Researchers debate what evidence would even show it
paulnovosad · x · 2026-10-28
Arpit Rage and Paul Novosad debate whether LLMs have internal states mappable to sadness or happiness. Rage notes LLMs make worse, more misaligned decisions when pushed into negative affect, and these states seem to persist for some time. Novosad challenges what evidence could support that, and Rage concedes it's hard to distinguish concepts emerging from human-language prevalence from something that genuinely resembles feeling.
Related event: Economists and researchers debate whether LLMs can suffer(5 posts)→
More from AGI Musings
- Drug development timelines stretch to 9.3 years as pipelines slow 32% despite AI advances — paulnovosad · 2026-10-28
- Committee reportedly holds back OpenAI's major math breakthrough over caution — IgorCarron · 2026-10-28
- Marc Andreessen: Europe is "determined to be left behind" on AI and would happily skip it entirely — 2C_ornot2C · 2026-10-28
- AI can prove famous math problems now, but open-ended science is the real test — miniapeur · 2026-10-28
- The Sane Asylum: Inside Silicon Valley's AI-Era Collective Breakdown — DevinLewtan · 2026-10-28
- Paras Chopra: "AI-generated" labels will fade; "100% human" will flourish — paraschopra · 2026-10-28