arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2304.09848cs.CLcs.IR
收藏

评估生成式搜索引擎的可验证性

Evaluating Verifiability in Generative Search Engines

Stanford University(斯坦福大学)

Nelson F. Liu, Tianyi Zhang, Percy Liang

AI总结:

本文通过人工评估审计了四种主流生成式搜索引擎,发现其响应中存在大量不受支持的陈述和不准确的引用,揭示了现有系统在可验证性方面的严重不足。

摘要:

生成式搜索引擎直接生成对用户查询的响应,并带有内联引用。一个可信赖的生成式搜索引擎的先决特征是可验证性,即系统应全面引用(高引用召回率;所有陈述均完全由引用支持)且准确引用(高引用精确率;每条引用均支持其关联的陈述)。我们进行了人工评估,以审计四种流行的生成式搜索引擎——Bing Chat、NeevaAI、perplexity.ai 和 YouChat——涵盖了来自各种来源(例如,历史 Google 用户查询、Reddit 上动态收集的开放性问题等)的多样化查询集。我们发现,现有生成式搜索引擎的响应流畅且看似信息丰富,但经常包含不受支持的陈述和不准确的引用:平均而言,仅有 51.5% 的生成句子完全由引用支持,且只有 74.5% 的引用支持其关联的句子。我们认为,对于可能作为用户信息检索主要工具的系统而言,这些结果低得令人担忧,尤其是考虑到它们表面上具有可信度。我们希望这些结果能进一步推动可信赖生成式搜索引擎的发展,并帮助研究人员和用户更好地了解现有商业系统的缺陷。

英文摘要:

Generative search engines directly generate responses to user queries, along with in-line citations. A prerequisite trait of a trustworthy generative search engine is verifiability, i.e., systems should cite comprehensively (high citation recall; all statements are fully supported by citations) and accurately (high citation precision; every cite supports its associated statement). We conduct human evaluation to audit four popular generative search engines -- Bing Chat, NeevaAI, perplexity.ai, and YouChat -- across a diverse set of queries from a variety of sources (e.g., historical Google user queries, dynamically-collected open-ended questions on Reddit, etc.). We find that responses from existing generative search engines are fluent and appear informative, but frequently contain unsupported statements and inaccurate citations: on average, a mere 51.5% of generated sentences are fully supported by citations and only 74.5% of citations support their associated sentence. We believe that these results are concerningly low for systems that may serve as a primary tool for information-seeking users, especially given their facade of trustworthiness. We hope that our results further motivate the development of trustworthy generative search engines and help researchers and users better understand the shortcomings of existing commercial systems.

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。

↑