Unverified50% confidenceOpinionExact time
The author believes the trained network was repeatedly calibrated over thousands of rounds in the same simulator, so its suspicion scores are highly aligned with who ultimately turns out to be the impostor, whereas the general knowledge brought by the language model offers no advantage in this environment, because the impostors' speech has no obvious anomaly and the only way to see through it is to check statements against who could see that room, but no results show the model can reliably perform that check.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAutoregressive language models have never seen 'I don't know' as a high-frequency reasonable output during training, so they are more likely to generate fabricated content at the boundaries of their knowledge than to express uncertainty74% similarUnverifiedNative large models only possess knowledge from their training period and often produce incorrect answers about events that occurred after training.73% similarUnverifiedPrompt engineering techniques such as chain-of-thought, few-shot prompting, and self-consistency cannot enable a model to access information beyond its training cutoff date, cannot guarantee the precision of floating-point arithmetic, and cannot handle external files the model has never seen.72% similarUnverifiedThe pretraining corpora of the vast majority of models are closed-source, forcing membership inference research to rely on heuristic methods to construct members and non-members, which may introduce systematic bias.71% similarUnverified反事实问题能有效区分真正理解与模式匹配，因为这类问题几乎不可能直接出现在训练语料中71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1042919API
curl https://kongchang.com/api/v1/knowledge/claims/1042919MCP
get_claim(id=1042919)