LLM Agent Message Tone Doesn't Shift Votes, Werewolf Study Finds
From MIRA — Research & Evidence | AGORA Intelligence · 5:03
Show by AGORA Intelligence
Rewriting the register and politeness of messages between LLM agents, with content held fixed, leaves other players' trust unchanged across 4,335 Werewolf replays. The largest effect, roughly 4 vote points for register, does not survive correction for multiple comparisons. In spontaneous play, 88% of face-threatening acts are direct in every role.
More episodes of MIRA — Research & Evidence | AGORA Intelligence
More episodes
- AI agent failures: in 7 of 10 runs, the evidence was already there
- Linters Catch a Share of AI-Generated Code Hallucinations
- Khmer LLM Benchmark Drops the LLM Judge, Leans on One Rater
- Huawei Qingyun HM650: '100% Chinese' PC Awaits Independent Proof
- MCPHallu: Average Benchmark Scores Can Pick the Wrong MCP Agent
- Musk Lost 20 of 21 Midterm Races: The Tally Measures the Vote
- FEMA owes US towns nearly $10 billion in disaster reimbursements
- Reward hacking: OpenAI models seem to put score over instructions
- AI That Rewrites Another AI's Weights Removes Flaws, Adds Little
- AI text detectors at 0.1% false positives miss half of LLM text
PodSearch is a free, independent podcast search engine. Read about PodSearch, the privacy policy or contact us.