Abstract Artificial Intelligence is involved in more and more fields, and the debate of whether human beings will be replaced by it is more and more intense. This paper explores the paradigm of ethical decision-making of artificial intelligence from the perspective of Marx’s philosophical thought. Based on the basic elements of ethical decision-making, three decision-making methods are proposed to complete the judgment of decision-making consequences, and the calculation formula for the consequences of these methods is derived. Using the Bayesian learning method, a utility-maximizing agent effect function is constructed to implement the moral embedding policy. Measure the ethical fairness of decision-making algorithms based on statistical fairness orientation to propose fair decision-making methods. Through the example analysis, the practical application effect of the model is analyzed from two perspectives: public policy implementation and education. More than 40% of the users are more satisfied with the public policy implementation made by AI, and they are relatively satisfied with the implementation of AI governance policy at this stage, but there is still room for improvement. There is a difference in the cognition of students of education-related majors and computer science majors on the ethics of identity intelligence, and education majors pay more attention to the educational level and social virtue issues, such as educational sovereignty, educational purpose, educational value, educational evaluation, educational fairness, as well as the development of students, and integrity issues.