Abstract
Large Language Models (LLMs) have shown a high capability in answering questions on a diverse range of topics. However, these models sometimes produce biased, ideologized or incorrect responses, limiting their applications if there is no clear understanding of which topics their answers can be trusted. In this research, we introduce a novel algorithm, named as GMRL-BD, designed to identify the untrustworthy boundaries (in terms of topics) of a given LLM, with black-box access to the LLM and under specific query constraints. Based on a general Knowledge Graph (KG) derived from Wikipedia, our algorithm incorporates with multiple reinforcement learning agents to efficiently identify topics (some nodes in KG) where the LLM is likely to generate biased answers. Our experiments demonstrated the efficiency of our algorithm, which can detect the untrustworthy boundary with just limited queries to the LLM. Additionally, we have released a new dataset containing popular LLMs including Llama2, Vicuna, Falcon, Qwen2, Gemma2 and Yi-1.5, along with labels indicating the topics on which each LLM is likely to be biased.
Abstract (translated)
大语言模型（LLMs）在回答各类话题的问题上展现出强大能力。然而，这些模型有时会产生带有偏见、意识形态化或不正确的回答，若无法明确判断哪些话题的答案可信，将限制其应用范围。本研究提出一种名为GMRL-BD的新型算法，旨在通过仅对LLM进行黑盒访问并在特定查询约束下，识别给定LLM在话题层面的不可信边界。该算法基于从维基百科导出的通用知识图谱（KG），结合多个强化学习智能体，高效识别LLM可能产生偏见回答的话题（KG中的部分节点）。实验证明，本算法效率显著，仅需有限次查询LLM即可检测其不可信边界。此外，我们发布了一个新数据集，涵盖Llama2、Vicuna、Falcon、Qwen2、Gemma2及Yi-1.5等主流LLM，并标注了各模型可能产生偏见的话题。
URL
https://arxiv.org/abs/2604.05483
