**DiscoBench基准测试揭示:AI搜索代理失败主因并非检索能力,而是缺乏澄清提问**

_AI search agents don&\#039;t fail at searching, they fail at asking the right questions when queries get ambiguous_

> 最新研究指出,AI搜索代理在多步研究中表现不佳的核心原因并非搜索本身,而是在面对模糊查询时未能主动请求澄清。DiscoBench基准测试显示,模型倾向于重复搜索而非追问,导致准确率仅51.9%,甚至低于随机猜测;而消除歧义后,准确率可提升高达40个百分点。

**来源信息**
- **来源**:The Decoder:AI News(RSS)
- **分类**:ai-models
- **发布时间**:2026-07-05 15:52(北京时间)
- **原文**:[打开原文](https://the-decoder.com/ai-search-agents-dont-fail-at-searching-they-fail-at-asking-the-right-questions-when-queries-get-ambiguous)