研究：多估计引导未能统一消除语言模型判断偏差
Multi-Estimate Elicitation Does Not Uniformly Debias Language Model Judgments
arXiv 上的预注册研究，用 gpt-6.1-sol 和 claude-sonnet-5.5 跑了锚定、后见之明等五类偏差任务，结论是让模型一次报五个估计并不能稳定去偏，做 LLM 评测的可以看看实验设计。
一项预注册研究在锚定、后见之明、基础率、合取等认知偏差任务上，比较单次回答中抽取五个加权估计与直接数值回答的效果。实验覆盖 gpt-6.1-sol、claude-sonnet-5.5 和 Qwen3-8B 三个配置，前两者在直接引导下无关锚点指数和现状效应接近零，同时仍对信息来源和建议保持敏感。23 项预注册验证性检验中有 6 项在 Holm 校正后拒绝原假设，但这 6 项的绝对效应量均低于预设的最小感兴趣效应。多估计引导并未在任务和模型间产生一致的去偏效果，候选答案离散度与锚定误差的事后相关性有待独立验证。
Multi-Estimate Elicitation Does Not Uniformly Debias Language Model Judgments
Does eliciting five weighted estimates within a single response change how a language model responds to information? Across anchoring, hindsight, base-rate, conjunction, and status-quo/default tasks, we compare direct numerical answers with a multi-estimate response format, contrasting source framings, explicitly random cues, and probability-consistency constraints. We evaluate three completed configurations--gpt-6.1-sol, claude-sonnet-5.5, and Qwen3-8B (reasoning disabled, analyzed descriptively). In the first two configurations, direct elicitation yielded near-zero irrelevant-anchor indices and status-quo effects, while remaining responsive to informative sources and informed advice. Of 23 preregistered confirmatory tests, six rejected after Holm adjustment; however, all six absolute point estimates fell below their prespecified smallest effect sizes of interest without establishing equivalence. Moreover, while separate versus joint elicitation shifted conjunction judgments, interpretation is constrained by between-call noise. Ultimately, multi-estimate elicitation failed to produce uniform debiasing or response attenuation across tasks and models, though post hoc correlations between candidate spread and anchoring error motivate independent predictive validation. Materials: https://github.com/asj252/llm-judgment-studies