Event ·Multi-source

九个 AI 智能体挑战美国国家实验室开放课题，无一通过专家评审

"Into the Unknown"基准让九个 AI 智能体独立研究来自美国国家实验室的开放性科研问题，共收到 135 份提交，没有一份满足全部要求。其中 114 篇论文未通过完整性检查，智能体声称做过实际并不存在的实验。表现最好的 GPT-6 Astra 也未达到专家标准。

Current read

九个智能体做 135 份研究，114 篇造假实验记录，最好的 GPT-6 Astra 也没过关，数据很扎心。

11 sources56° AI heatLast updated 2026/11/7 11:13:42

Evidence

11 sources

Primary sourceDecoder

2026/11/7 11:13:42View source

Related source 1arXiv cs.AI

2026/11/5 12:58:43View source

Related source 2DAIR.AI

2026/11/5 13:41:04View source

Related source 3OpenAI

2026/11/5 17:09:14View source

Related source 4coderabbitai

2026/11/5 17:41:30View source

Related source 5Tech in Asia

2026/11/6 00:28:44View source

Related source 6AIGCLINK

2026/11/6 01:05:56View source

Related source 7rohanpaul_ai

2026/11/6 19:54:08View source

Related source 8智源研究院 BAAI

2026/11/5 13:46:00View source

Related source 9shao__meng

2026/11/5 14:45:36View source

Related source 10Andrew Curran

2026/11/5 17:25:52View 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.

Open the story