arXiv · cs.AI · Xingyu Yi, Xinya Zhang, Yunshen Zhou, Haoyu Bao · 9 小时前模型评分41
BioAgent-DF 审计生物医学研究智能体的决策溯源
From Tool State to Scientific Action: Auditing Decision Provenance in Biomedical Research Agents
导读
BioAgent-DF 通过在同一科学计算位置保存并还原参考值与候选值，审计生物医学研究智能体的决策溯源。在八个 SciFact 批次中，Qwen3.5-4B 出现三组可复现的引用变化，Gemma 3 4B IT 出现四组可复现的读取与引用变化；完整还原可恢复参考证据集而结论不变。
分类置信度
中 · 待人工复核
- 只有单一媒体来源，需要人工复核。
来源级别：媒体与个人
应用领域：医疗与生命科学、科学研究
事件类型：研究发布
证据摘录
- “BioAgent-DF tests this separation using paired reference and candidate values saved at the same scientific computation site.”
- “Full restoration recovers the reference evidence sets while verdicts remain unchanged.”
- “After the agent had used faulty values, restoring its earlier execution state and supplying certified scores recovered all 39 changed decisions, whereas replacing the prediction file recovered none; no incorrect action was committed.”
来源：arXiv · cs.AI · arxiv.org