arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2405.14812cs.CY
收藏

作为AI语言模型,“是的,我会建议报警”:大语言模型决策中的规范不一致性

As an AI Language Model, "Yes I Would Recommend Calling the Police": Norm Inconsistency in LLM Decision-Making

Shomik Jain, D Calacci, Ashia Wilson

AI总结:

本研究针对LLM在监控场景下的报警决策展开,评估三款前沿模型发现其存在显著规范不一致性,存在与犯罪事实脱节及受社区种族构成影响的偏差,凸显相关偏见应对策略的局限。

摘要:

我们研究了规范不一致现象,即大语言模型(LLM)在相似情境下应用不同规范的情况。具体而言,我们聚焦于高风险应用场景:针对Amazon Ring家庭监控视频判断是否应当报警。我们评估了三款前沿LLM——GPT-4、Gemini 1.0和Claude 3 Sonnet——的决策,考察维度包括视频呈现的活动、受试者的肤色与性别,以及视频录制社区的特征。我们的分析揭示了显著的规范不一致性:(1)报警建议与实际犯罪活动存在与否之间存在脱节;(2)决策偏差受社区的种族人口构成影响。这些结果凸显了监控场景下模型决策的随意性,以及当前偏见检测与缓解策略在规范性决策中的局限性。

英文摘要:

We investigate the phenomenon of norm inconsistency: where LLMs apply different norms in similar situations. Specifically, we focus on the high-risk application of deciding whether to call the police in Amazon Ring home surveillance videos. We evaluate the decisions of three state-of-the-art LLMs -- GPT-4, Gemini 1.0, and Claude 3 Sonnet -- in relation to the activities portrayed in the videos, the subjects' skin-tone and gender, and the characteristics of the neighborhoods where the videos were recorded. Our analysis reveals significant norm inconsistencies: (1) a discordance between the recommendation to call the police and the actual presence of criminal activity, and (2) biases influenced by the racial demographics of the neighborhoods. These results highlight the arbitrariness of model decisions in the surveillance context and the limitations of current bias detection and mitigation strategies in normative decision-making.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑