Is Deep Research Reliable? Misleading Knowledge Induces False Conclusions

  • 类型:arxiv
  • 标识:2607.20891
  • 链接:https://arxiv.org/abs/2607.20891
  • 主分类:agent
  • 形态:method
  • 被引:1
  • 被引来源:Semantic Scholar
  • S2被引:1
  • OpenAlex被引:0
  • 影响力被引:0
  • TLDR:MisKnow-Agent is introduced, a controlled evaluation framework that constructs task-specific documents supporting manually audited false conclusions with controlled authority cues and source styles that evaluates DeerFlow and WebThinker with three backbone LLMs using a report-level false-conclusion adoption rate that counts only reports endorsing the false conclusion.
  • OpenAlex ID:W7170676285
  • OpenAlex DOI:10.48550/arxiv.2607.20891
  • DOI:10.48550/arxiv.2607.20891
  • DOI来源:OpenAlex
  • 开放获取:green
  • 开放获取链接:https://doi.org/10.48550/arxiv.2607.20891
  • OpenAlex更新:2026-08-26
  • 待LLM分类:否
  • 标题中文:Deep Research 可靠吗?误导性知识会诱发错误结论
  • TLDR中文:提出 MisKnow-Agent,一个受控评测框架,通过可控的权威线索与来源风格构造支撑人工审核结论的任务文档,并采用报告级 false-conclusion 采纳率(仅统计认可错误结论的报告),基于三种 backbone LLM 评估 DeerFlow 与 WebThinker。
  • 来源文件
  • /inbox/tom/_candidates/2026-07-31-agent-rag-longcontext-candidates.json
  • [S2 enrich]
  • [OpenAlex backfill]