INTEGRATION OF ARTIFICIAL INTELLIGENCE INTO STEM EDUCATION FOR PREDICTING STUDENT PERFORMANCE AND ANALYZING ETHICAL CHALLENGES

Abstract

The article is devoted to the challenges of integrating artificial intelligence (AI) into STEM education, particularly in the context of the courses “Fundamentals of Robotics” and “STEM Education and Robotics” for students of Berdyansk State Pedagogical University, as well as to methods of addressing these challenges through modern technologies. The current state of AI development in education is examined, with a focus on learning analytics, which enables the prediction of students’ academic performance based on data from learning management systems such as Moodle. The components of the courses are discussed, including programming Arduino microcontrollers to control potentiometers, photoresistors, servomotors, RGB LEDs, ultrasonic sensors, and membrane keyboards, as well as ethical challenges related to AI use, particularly regarding data privacy and academic integrity. Based on the analysis, the problem of insufficient use of AI for performance prediction in robotics courses and the lack of ethical standards in Ukrainian education is identified, highlighting the need to improve the quality of student training through the implementation of modern learning analytics methods and harmonization with European standards. The object of the study is defined as the professional training of future computer science teachers in STEM education, and its purpose is to develop a model for integrating AI to predict academic performance and analyze ethical challenges. A set of research tasks, a general hypothesis, and several partial hypotheses are formulated. The general hypothesis states that the developed model of AI integration into STEM education will be effective if qualitative changes are achieved in students’ knowledge of learning analytics and Arduino programming principles; if conditions are created for students’ self-realization through practical laboratory work; and if the course content emphasizes mastering the skills of using AI for data analysis and forecasting. It is proven that students should have the opportunity to receive a comprehensive education that includes theoretical training in programming fundamentals, electronics, and AI ethics, as well as practical skills in working with Arduino microcontrollers. The use of LMS data to predict the academic performance of 62 second-year students of the Faculty of Physics, Mathematics, Computer Science, and Technology Education is described, including login frequency, task completion time, and the number of completed laboratory assignments. The results of prediction using linear regression and clustering are presented, which made it possible to identify students at risk of failure and adapt the educational process. Ethical challenges such as data protection and plagiarism prevention are outlined, and recommendations are proposed for harmonizing Ukrainian practices with European standards. The problem for further research is formulated as the need to improve AI models for complex programming tasks and to develop ethical standards for their use in education.