Event ·Multi-source

研究：Claude Opus 4.8 在同行评审中表现匹敌人类专家

一项研究用 124 篇哈佛社会科学论文测试 Claude Opus 4.8 的同行评审能力，覆盖统计分析、形式化模型和定性研究等多种类型。结果显示 Opus 4.8 在各项指标上都达到人类专家评审员的水平。Ethan Mollick 转发了这项研究，并指出学术界需要尽快调整科研实践来适应 AI。

Current read

有人拿 124 篇哈佛论文测试 Opus 4.8 当审稿人，结果和人类专家打平，搞科研的朋友可以看看细节。

11 sources61° AI heatLast updated 2026/10/23 04:10:04

Evidence

11 sources

Primary sourceEthan Mollick

2026/10/23 04:10:04View source

Related source 1DAIR.AI

2026/10/22 14:24:02View source

Related source 2swyx (AI Engineer)

2026/10/21 05:06:02View source

Related source 3TestingCatalog

2026/10/21 19:15:11View source

Related source 4kimmonismus

2026/10/21 19:47:25View source

Related source 5rohanpaul_ai

2026/10/21 19:48:33View source

Related source 6IT之家

2026/10/21 22:44:18View source

Related source 7AIGCLINK

2026/10/21 23:20:38View source

Related source 8Gorden Sun

2026/10/21 23:36:20View source

Related source 9岚叔

2026/10/22 00:24:37View source

Related source 10Artificial Analysis

2026/10/22 02:04:56View 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