人工智能的道德判断及其伦理建议

学术界关于AI的道德判断主要存在三种不同的立场与观点。然而，对AI进行道德判断需要区分两个问题：其一是AI本身的道德评价问题；其二是AI研发与应用后果的善恶评价问题。二者既存在差别，又紧密相联。前者解决的关键在于AI道德判断的前提性规定，即AI是否真正意义上的人格实体与道德主体问题的讨论，而后者解决的关键仍在于人类自身。只有在人类与AI的关系等进行批判性反思的基础上才能对上述问题作出道德判断。具体而言，当前对AI应采取以下审慎态度：一是在大力研发和使用“弱AI”同时，应适当限制“强AI和超强AI”的研发和使用，尤其应当限制甚至禁止“杀人机器人”的研发与使用；二是国际社会、行业协会等应为不同发展阶段AI的研发与应用制定公约，确立行业发展伦理标准，并敦促相关法律法规出台，通过道德调整与法律规制引导AI健康发展；三是通过道德教育培养科技人员的道德自觉意识与社会责任意识。科技人员应当遵守科技伦理与职业道德，履行“通告和预防义务”
There are mainly three types of understanding of the moral judgement on artificial intelligence (AI).In this paper, we argue that the moral judgement of AI should be based on the distinction of two issues, i.e.the judgement on AI itself and the evaluation of AI development and application. These are two independentissues but at the same time they are closely connected with each other: The former concerns the premise ofmoral judgement on AI, i.e. whether or not AI is an entity with personality and a morality subject in a realsense; the latter’ s solution, however, depends totally on the human beings themselves. Moral judgments can bemade only on the critical reflection on the relations between human beings and AI systematically. Specifically,prudent attitudes towards AI should be taken. First,‘ weak AI’ should be developed and applied vigorously,while‘ strong AI’ and‘ super strong AI’ should be restricted, and in particular the development of‘ humankilling robots’ should be restricted and prohibited. Second, international communities and industry associationsshould formulate norms guiding the development and application of AI at different stages, establish ethicalstandards and relevant laws and regulations governing the development of AI, hence a healthy developmentof AI. Third, the technicians should cultivate the consciousness of morality and social responsibility, followthe scientific and technological ethics and professional ethics, and carry out the obligation of‘ notification andprecaution.