Informal algorithm auditing
由非专业用户对 AI 系统进行系统性较弱、更具探索性的审计：通过反复用不同输入试探系统并观察输出，揭示潜在偏见或危害。与专家审计（如 Buolamwini & Gebru 的性别分类审计）相对。
English Definition
"A less systematic, exploratory form of algorithm auditing in which non-expert users query an AI system with varied inputs and observe outputs to surface harmful or biased behaviors."
用法说明 · 针对中文母语者
对比 expert audit：informal 算法审计通常更依赖日常用户的探索而非严谨方法论；近年 HCI 文献将其视为推动公众参与 AI 治理的途径。
真实用例 · 3 条
"We draw from prior work on participatory informal algorithm auditing (Sec. 2.3), in which non-expert participants explore and evaluate the in/accuracy of varied inputs and corresponding outputs of AI or other algorithms."
"Informal auditing is typically less systematic and more exploratory."
"In addition to expert audits such as Buolamwini and Gebru's audit uncovering intersectional racist and sexist bias in gender classification [23], informal audits by everyday users can uncover harmful algorithmic behaviors [72, 119]."