Abstract: In the context of the digital and intelligent technologies empowering the modernization of education, artificial intelligence technology breaks through the limitations of time and space to build an adaptive learning platform for teachers and students. However, ethical awareness lags behind technological innovation, which has led to new development bottleneck of artificial intelligence in education. From the perspective of Simon's theory of bounded rationality, an ethical reflection on the application of artificial intelligence in education reveals thatthere are still a series of ethical dilemmas, such as weakened subjectivity of teachers and students, data distortion, privacy leakage, lack of, information security failure, and regional education inequality. The bounded rationality of the three behavioral subjects-teachers, students and technical personnel—as well as the constraints of internal and external environments are the key factors leading to these dilemmas. Therefore, it is necessary to strengthen the construction of the ethical governance system for artificial intelligence to overcome algorithmic biases, go beyond instrumental rationality to improve the ethical framework for AI applications in education, adhere to the leading role of teachers and the subject status of students to return to the essence of education, in order to break through the dilemma of bounded rationality of both behavior subjects and the environment, and promote the high-quality development of education empowered by artificial intelligence.