Ethical Governance of AI:Basic Rationale,Current Strategies and a Practical Turn
The basic rationale for the ethical governance of AI is derived from foundational theories in eth-ics:teleology,deontology,and virtue ethics.These frameworks either focus on the outcomes of AI ac-tions,the principles guiding AI actions,or the moral cultivation of AI.From the perspective of AI exer-ting ethical functions,this can be categorized into four levels:ethical agents,implicit ethical agents,ex-plicit ethical agents,and complete ethical agents.Based on this fundamental rationale,the current strate-gies for ethical governance involve top-down,bottom-up,or hybrid approaches to enable AI to learn the basic principles of moral decision-making.The methods for this acquisition involve using deontological logic,cognitive logic,action logic,and deontological-cognitive-action logic to translate ethical princi-ples into algorithms that AI can recognize and accept.However,a persistent issue within these current strategies is whether ethics can truly be computationalized.Therefore,a practical turn is required,whereby humans and AI form a morally symbiotic relationship,leveraging their complementary strengths in specific moral practices.
NETL
NSTL
万方数据
