Unverified50% confidenceFactExact time
The evaluation uses two AI judges: Anthropic's Claude Opus 5.5 and OpenAI's GPT-5.6 Sol, who cannot see each other's judgments, with bias eliminated through anonymizing and randomly ordering answers, a second round of letter-reordering review, and setting up control tests.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/13/2026
First Seen
Valid until: 1/11/2027
Sources
Related Entities
Related Claims
UnverifiedAn independent advisory group of 9 scholars (including 3 Fields Medal winners) issued a statement declining to endorse OpenAI's mathematical results or the process used to produce them.66% similarUnverified主流AI模型存在谄媚性倾向，只需对模型说一句「你确定吗」便能让其推翻原本正确的答案，测试覆盖GPT-5.5等模型65% similarUnverifiedThe test conclusions come from a single tester focused on the specific scenario of WeChat Official Account typography replication; the scores are subjective and do not represent the models' full capabilities in dimensions such as programming, reasoning, or long-context handling.64% similarUnverifiedThe manual distinguishes between two situations in which confidence in the correct answer declines: simply losing its edge versus being actively suppressed by fine-tuning.63% similarUnverifiedTakedown decisions based on AI often lack transparent and efficient appeal channels.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1016224API
curl https://kongchang.com/api/v1/knowledge/claims/1016224MCP
get_claim(id=1016224)