角色引导不能真正消除大模型偏见，只改变输出表层表现。
The Illusion of Debiasing: Persona Steering Redistributes Rather Than Reduces Bias in LLMs
- 用角色设定探测模型内部结构变化，发现仅影响输出层面
- 角色指令无法还原人类特质间的关联性，偏见依然存在
- 越深层任务中，角色引导对偏见影响越弱，表面有效实则无效
基于提示的干预（如系统提示、角色设定、角色指令）能可靠地改变语言模型的输出，但其作用层级尚不明确：是重构模型内部结构，还是仅调节输出通道？我们以角色条件作为可控探针，沿从自述、开放生成到词级参数关联的深度轴线，在三个指令微调模型上测量其影响。结果呈现渐进式分离：角色设定可被识别但非结构性；模型虽能遵循单一特质指令，却无法再现人类特质间的共变关系。这种分离随深度加深：角色设定维持或加剧封闭式问答中的偏见，改变整体语气但未缩小群体间差异，且几乎不扰动已饱和的关联基线。因此，提示引导仅作用于输出通道，其结构性影响有限，而表面可操控性可能掩盖这一局限。
原文摘要 · Abstract (English)
Prompt-based interventions: system prompts, personas, role instructions, reliably reshape what a language model says, but it is unclear which layer they reach. Do they reconfigure internal structure, or only modulate the output channel? We use persona conditioning as a controlled probe, measuring its effects along a depth axis from self-report, through open-ended generation, to word-level parametric association, across three instruction-tuned models. We find a graded dissociation. Personas are legible but not structural: models follow single-trait instructions yet fail to reproduce human inter-trait covariance. The dissociation deepens with depth: personas hold or amplify closed-form QA bias, shift absolute tone while leaving between-group disparity unchanged, and barely perturb an already saturated associative baseline. Prompt-based steering thus operates in the output channel and has a structural reach limit that surface manipulability can mask.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。