Event ·Multi-source

AI模型更倾向于选择自己的答案

研究人员比较了12个模型在1,460场对决中的34,580次判断结果。AI模型平均选择自己答案的频率为58%,而人类仅为34%。GPT-6 Astra选择自己的比例高达88%。AI模型之间的一致性达到79%,而与人类的一致性仅为57%。GPT-5.6 Sol在96%的情况下都选择胜者,而人类有32%的情况选择平局或两者都不好。

Current read

Arena研究发现AI模型有自己独特的判断偏好,与人类明显不同,GPT-6 Astra甚至88%选择自己。

11 sources73° AI heatLast updated 2026/9/29 16:19:03

Evidence

11 sources
Related source 3arXiv cs.AI
View source
Related source 4Perplexity
View source
Related source 5kimmonismus
View source
Related source 6Andrew Curran
View source
Related source 8OpenAI Blog
View source
Related source 9The Rundown AI
View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

AI模型更倾向于选择自己的答案 · 事件专题 | AITOP