跨引擎AI搜索引用差异显著：无引擎评分难以预测可见性
原标题：Scoring With the Engine: Retrieval Exposure, Cross-Engine Divergence, and the Limits of Engine-Agnostic GEO Scores
AI 摘要
该研究通过2026年6月6日对ChatGPT、Microsoft Copilot、Google和Perplexity四个AI搜索界面的观测审计，用15个商业提示词收集了589条引用观测，发现同一提示词下跨引擎的URL重叠极低（平均Jaccard相似度仅0.0079，84.9%的引擎对无共享URL），96.4%的URL只出现在一个引擎中。研究指出，不依赖引擎的页面评分只能估计页面质量或查询-页面匹配度，端到端可见性还取决于引擎特有的曝光与选择过程，因此不应将此类'AI可见性分数'直接解释为被引用概率。
正文节选
Scoring With the Engine: Retrieval Exposure, Cross-Engine Divergence, and the Limits of Engine-Agnostic GEO Scores Abstract Recent work asks whether generative-engine visibility can be approximated with deterministic, engine-free page scores. Bajemon and Rochet (2026) provides an unusually careful answer: query-agnostic page scores have weak within-query association with citation order, while query-conditioned relevance is substantially more informative in controlled candidate-set experiments. T