世界卫生组织于2026年6月2日发布讨论文件，系统分析人工智能在循证健康政策制定中的应用机会与潜在风险。文件关注的范围不只包括临床应用，还延伸到政策问题界定、证据整理、方案设计和实施效果评估等完整政策周期。
人工智能能够提高信息检索、证据综合和数据分析效率，但如果训练数据、模型方法和使用过程缺乏透明度，也可能带来偏差、错误放大和责任界定不清等问题。政策制定机构需要能够理解工具的适用边界，并保留必要的人工判断与复核机制。
讨论文件梳理了循证决策工具与人工智能治理框架之间的共同原则，包括透明度、公众参与、权利保护和基于风险的监督。其核心不是另起一套孤立规则，而是把人工智能治理要求嵌入现有政策制定和证据评估体系。
对于医学、生命科学和公共政策领域的研究者而言，这一趋势意味着技术能力与证据质量、伦理审查和治理能力需要同步建设。
信息来源：世界卫生组织（WHO）
On 2 June 2026, the World Health Organization published a discussion paper examining the opportunities and potential risks of using artificial intelligence in evidence-informed health policy. Its scope extends beyond clinical applications to the full policy cycle, including problem definition, evidence synthesis, policy design, and impact assessment.
AI can improve the speed of information retrieval, evidence synthesis, and data analysis. However, opaque training data, methods, or deployment processes can also introduce bias, amplify errors, and make accountability difficult. Policy institutions therefore need to understand the limits of each tool and retain appropriate human judgement and review.
The paper identifies shared principles between evidence-informed policy tools and AI governance frameworks, including transparency, participatory engagement, rights protection, and risk-based oversight. The aim is to integrate AI governance into existing policy and evidence processes rather than create an isolated set of rules.
For researchers in medicine, life sciences, and public policy, this development reinforces the need to build technical capability together with evidence quality, ethics review, and governance capacity.
Source: World Health Organization