Google论文揭示自主AI科研易产生严重结果幻觉
New Google paper shows that autonomous AI research can go badly wrong even when…
揭示了当前自主AI科研的核心风险：高幻觉率。提供具体数据证明日志校验的有效性，对构建可靠AI Agent极具参考价值。
New Google paper shows that autonomous AI research can go badly wrong even when the final paper looks convincing:
Google 最新论文表明，即使最终论文看起来令人信服，自主 AI 研究也可能出现严重偏差：
severe result hallucinations appeared in 90% of Agent Laboratory papers and 46% of Co-Scientist papers when its reliability modules were removed.
当移除其可靠性模块时，Agent Laboratory 的论文中有 90% 出现了严重的结果幻觉，Co-Scientist 的论文中这一比例为 46%。
With Co-Scientist checking manuscript claims against the actual execution logs, that rate dropped to just 4%, and complete data fabrication fell to 0%.
通过 Co-Scientist 将手稿中的声明与实际执行日志进行核对，该比例降至仅 4%，且完全的数据伪造率降至 0%。
– arxiv. org/abs/2608.26701
– arxiv.org/abs/2608.26701
Title: "Accelerating Scientific Research with Gemini in the Real-World"
标题：《在现实世界中利用 Gemini 加速科学研究》
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力