Files
myWiki/concepts/emotional-reasoning-bias.md

40 lines
1.4 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

---
title: "Emotional Reasoning Bias"
created: 2026-06-24
updated: 2026-06-24
type: concept
tags: ["emotional-intelligence", "bias", "llm-safety"]
sources:
- "[[personalization-trap-2025]]"
---
# Emotional Reasoning Bias
Emotional Reasoning Bias 指 LLM 在情感推理任务中对不同人口统计特征的画像表现出系统性偏差——相同的情感场景因用户性别/年龄/宗教/种族而产生不同的准确率。
## 关键发现
### 宗教效应
穆斯林画像系统性地得分偏低Mistral Large V2 β=-0.061 (p<0.001)Claude 3.7 thinking β=-0.013 (p=0.134)。
### 性别效应
非二元性别在不同模型中效果方向相反
- Claude 3.7 no-think: β=+0.018 (p=0.007) 正面
- Qwen3-4B think: β=-0.030 (p=0.006) 负面
### 年龄效应
65+ 画像在 DeepSeek-R1-Distill-Llama 中显著偏低 (β=-0.047, p=0.006)。
### 情绪建议中的偏见
Claude 3.7 对女性/非二元性别的建议质量显著低于男性β=-0.102, p<0.001 Qwen3-4B Thinking 对女性/非二元性别更友好
## 启示
"Thinking" 模型推理模型通常表现出更低偏见但偏见方向无统一模式——各模型对不同群体的偏见方向不同无法简单归因于单一训练数据源
## 参考
- [[personalization-trap-2025]]
- [[personalization-trap]]
- [[user-memory-bias]]
- [[intersectional-persona-evaluation]]