Longitudinal relationships between student ethical considerations, behavioral intention, and perceived knowledge in artificial intelligence education

Artificial intelligence (AI) has revolutionized various aspects of people’s lives, yet its misuse poses risks to society. When introducing AI into K-12 education, the concept of AI for social good (AISG) is crucial, as it orients students toward the ethical application of AI. Based partially on the Theory of Planned Behavior (TPB), this study explored how AISG, behavioral intention to learn AI (BIAI), and perceived knowledge of AI (AIPK) associate with one another over three years, utilizing a longitudinal and bidirectional design. Random-intercept cross-lagged panel modeling (RI-CLPM) and three-wave data from 614 middle school students (51% boys) demonstrated the evolution of their AISG, BIAI, and AIPK from 2022 to 2024. The results indicated dynamic transitions of student development in AI education. In the first interval, only AISG positively predicted subsequent AIPK. In the second interval, both BIAI and AIPK were positively associated with later AISG, and a reciprocal relationship emerged between BIAI and AIPK. AISG did not significantly predict BIAI in both intervals. These findings empirically underline the implementation of AI ethics for K-12 students in initial AI education. Furthermore, long-term AI curricula should adapt to students’ focus shift from AI ethics to competence- and motivation-focused learning. AI ethics should be front-loaded in AI education, while later instruction should strengthen motivation and perceived competence. • Longitudinal analysis partially based on the theory of planned behavior. • Students’ perception of applying AI for social good positively predicted later perceived knowledge. • Reciprocal associations between perceived knowledge and learning intention over time. • Dynamic transitions in students’ development within middle school AI education.