In the deepening educational changes, personalized education has become a topic of great concern. The study mainly explores the use of artificial intelligence technology to construct a personalized learning environment and uses the assessment of learning performance in psychological education as an entry point for specific analysis. Comparative experiments are conducted to test the feasibility of improving the standard ant colony clustering algorithm to construct a model for psychological education performance assessment. The personalized learning system is then constructed with the performance assessment model, and students majoring in psychological education in a certain school are selected to conduct teaching experiments to explore the influence of the personalized learning system based on the improved ant colony clustering assessment model on psychological education. The grades of psychological education classes classified using the model of this paper are normally distributed, and it is reasonable and objective to use dynamic student achievement clustering to evaluate their grades. After the teaching experiment, students in the practice group were overall 2.98% to 6.72% higher than students in the other group in psychoeducational learning achievement, technology acceptance, and satisfaction, and 28.14% lower in learning anxiety. The personalized learning system designed in the study was able to improve students’ psychoeducational learning performance and experience, and reduce their learning anxiety, resulting in positive experimental results.