YouGov以真人逐一验证AI孪生
AI 总结
YouGov与OriginalVoices依据经身份验证的受访者创建了1,008个基于LLM的“E-twins”,并在48小时内通过回访向其中693人提出相同的新问题。YouGov称,加入受访者对4,000多个品牌和实体的评分后,孪生与对应真人在五分量表上误差不超过一分的逐人一致率从78%升至90%。在未纳入训练数据的虚构书籍主题上,孪生的一致率同样达到90%;另有75%的受访者认为孪生的开放式答案反映了自身观点。YouGov呼吁模拟服务商逐一验证孪生与对应真人的结果、测试未见主题、公布方法与失误,并公开预测现实结果。
深度见解
这项研究没有只比较总体结果,而是将每个AI孪生与其对应的受访者逐一比较,为检验个体模拟能否反映真人态度提供了一种方法。YouGov还指出,公开数据集可能已进入语言模型的训练数据,而总体比较可能掩盖趋向平均值或刻板化的孪生,因此行业需要共同验证标准。

New York City, United States, November 10th, 2026, TechnologyWire
- YouGov created 1,008 “E-twins” and tested them against 693 real, government-ID verified people they were built from, asking both the same new questions
- Twins’ results were validated through a recontact survey of the panelists they were based on within 48 hours
- Twins trained on panelists’ ratings of 4,000+ brands and entities matched their human to within one point on a five-point scale 90% of the time, up from 78% on profiling data alone
- Twins hit the same 90% on a topic they were never trained on
- YouGov calls on the industry to agree on validation standards, and challenges every simulator to predict real-world outcomes in public