人工智能，是指可以可能像人类一样、能够理性思考和行动的智能系统。严格来讲，目前的人工智能没有意图（Intention），因此实际上并不是根本意义上的人工智能。但现今，各方约定俗成，普遍将现在的技术发展阶段阐述为弱人工智能发展阶段。随着人工智能技术的兴起，其在医学领域广泛应用。相较于传统医疗服务，人工智能医学应用可提高医疗质量，降低医疗成本，优化治疗方案，提高医疗效率。

       然而，人工智能医学应用在我国面临若干伦理挑战。人工智能医学应用在实践过程中可能产生安全性问题。当它对人造成伤害时，面临严峻的责任归属问题。人工智能医学应用的基本要素之一是医疗健康数据。这些数据因有意、无意泄露、或遭受网络攻击，最终导致隐私泄露。而人工智能医学应用的算法偏倚会增强对个人和群体的歧视。一些人工智能医学应用的治疗价格高于传统医疗手段，那么让公众享受得到且负担得起相关服务是一个巨大挑战。此外，人工智能医学应用面临可解释性和透明性问题，它们可能导致公众的不信任。当医生建议/决定利用人工智能医学应用时，这种人机之间的不信任会使公众和病人怀疑医生是否为病人的利益做出最佳的判断，从而影响医患关系。

       因此，我国亟需采取有力的伦理措施解决人工智能医学应用的伦理问题，即提高各方利益相关者的伦理意识、建立健全人工智能医学应用的法律法规、监管标准、行业规范和公司准则，鼓励多方参与制定有关政策，加强相关培训和国际合作。

       本文首先梳理了人工智能技术及其医学应用的科学背景，结合我国的实际情况探讨人工智能带来的歧视、隐私、可解释性、透明性和信任、安全性、可及性和可负担性、责任归属这六大伦理和法律问题。其次建立评价人工智能医学应用的伦理框架，并对上述伦理问题进行伦理辨析。再次梳理欧洲和美国的伦理治理政策，剖析我国的伦理治理现状。最后基于我国国情和国外领先伦理治理经验，针对我国的法律制定者、政策制定者、专业共同体、科研机构、医疗卫生机构、行业和公司提出伦理治理建议。

       Artificial intelligence (AI) refers to an intelligent system that can think and act rationally like human beings. With the development of science and technology, AI technology has been widely applied in medical and healthcare fields. Compared with traditional medical services, AI applications in medicine can improve medical quality, reduce medical costs, optimize treatment options, and increase medical efficiency.

       However, AI applications in medicine pose ethical challenges in China. When an application regarding life and health is put into use, we firstly consider whether it is safe. It is obvious that AI applications in medicine may create safety issues. One of the basic elements of AI applications in medicine is healthcare big data, which relates to privacy and data security. The data may be transferred to a third party, or the cloud storage, or may be attacked, which may lead to privacy leakage. AI applications in medicine may enhance discrimination. Some AI applications in medicine are much more expensive than traditional medical services. Hence, promoting AI applications in medicine more accessible and affordable is a huge challenge. Besides, we should pay attention to the distributive justice of AI applications in medicine, and endeavor to find a reasonable solution. In addition, some AI applications in medicine are unexplainable or opaque, which may cause serious public distrust issues.

     This paper firstly introduces the scientific background of artificial intelligence technology and its applications in medicine, and introduces the ethical challenges. Then it builds an ethical framework of judging medical AI applications, and discusses ethical issues in detail, including discrimination, privacy, explainability, transparency and trust, safety, accessibility and affordability issues. Finally, it summarizes important international governance experience and China’s ethical governance status of AI applications in medicine, as well as puts forward the ethical governance suggestions to various stakeholders.