Recently, the use of artificial intelligence (AI) technologies has developed many aspects of education, as reported by several studies. However, little research has been done on the impact of AI-mediated education on junior middle school students' psycho-affective factors and behaviours. To fill the gaps, this study took an interventional mixed-methods research design to find out the contributions of AI tools to students' self-regulated learning (SRL). A sample of 78 Chinese junior middle school students completed a questionnaire on SRL before and after an intervention period powered by AI tools. They were assigned to control (n = 39) and experimental (n = 39) groups. The results of the t-test indicated a significant increase in the SRL mean score of the experimental group from pre-test to post-test, suggesting the efficacy of the intervention. Moreover, the results of thematic analysis, in the qualitative phase, revealed that AI technologies had contributed to Chinese junior middle school students' SRL by 'enhancing classroom engagement', 'providing personalised learning path', 'offering real-time and adaptive feedback', and 'encouraging an autonomous reflection on learning'. The findings are distinctly discussed, and implications are provided for the theory and practice of AI in junior middle school education. Directions are also provided for different stakeholders to encourage an AI-powered education that takes learner psychology into account.