N=815 study finds anthropomorphic language barely shifts public perceptions of AI, sparking debate
AndrewLampinen · x · 2026-10-11
Andrew Lampinen and ankit2119 clashed over whether anthropomorphic language for AI ("wanting", "persuading") misleads the public. Lampinen cited an arXiv paper by Betty Li Hou, Sophie Hao, Sunoo Park and Tal Linzen: an experiment with 815 participants comparing perceptions of AI after reading passages with vs. without anthropomorphic framing found the language style has little immediate effect on how people understand AI or assign responsibility — while a text explicitly discussing AI dangers did shift views, showing texts can move opinions but anthropomorphic wording itself isn't a major driver. Ankit argued the debate commits a fallacy: metaphor is fine generally, but anthropomorphic terms smuggle in misleading assumptions, and hindsight interpretations lack established causality. Lampinen challenged him for evidence that such metaphors are actually counterproductive.
More from AGI Musings
- 'Maths is over' is hubris: one mathematician says it has only just begun — timigod · 2026-10-11
- When AI can do the work, jobs shift from doing to judging what's worth doing — AryHHAry · 2026-10-11
- AI skeptic camp is self-contradictory: 'useless' yet career-threatening, argues creator — VraserX · 2026-10-11
- Survey: patients are nearing the tipping point of trusting AI more than their doctors — RishabJainK · 2026-10-11
- Only Anthropic told NYC lawmakers its models write code for smarter successors — rohanpaul_ai · 2026-10-11
- Satirical take paints AWS as a wrapper agent middleman users no longer need — max_paperclips · 2026-10-11