The first scene of artificial intelligence curriculum reform: longitudinal evidence on teachers' stress and...

The formal incorporation of artificial intelligence courses into basic education has raised a key issue for the effective implementation of the curriculum: whether teachers can withstand the continuously rising teaching pressure in the early phase of reform and prevent such pressure from further developing into teaching anxiety. Limited attention has been paid to the persistent influence of teachers' perceived AI teaching stress on teaching anxiety and the potential buffering role of school support in this process.

Drawing on two-wave longitudinal tracking data, this study analyzed 553 valid primary and secondary school teachers at T1 and 348 matched teachers at T2 to examine the persistent impact of perceived AI teaching stress on teaching anxiety, as well as the buffering effect of school support.

Teachers' stress and anxiety levels at T2 were both higher than those at T1, indicating an upward trend for both in the early stage of reform. Stress exerted a stable and significant positive effect on anxiety: stress at T1 significantly predicted anxiety at T1 (β ranging from 0.382 to 0.428, p < 0.001), and stress at T2 also significantly predicted anxiety at T2 (β ranging from 0.361 to 0.406, p < 0.001). Further analyses revealed clear cross-time continuity in both stress and anxiety, with stress at T1 significantly predicting stress at T2 (β ranging from 0.518 to 0.523, p = 0.001). Additionally, emotional support and training support significantly weakened the strength of the association between stress and anxiety.

The findings indicate that teacher anxiety in the early phase of artificial intelligence curriculum reform is not a short-term psychological fluctuation, but a sustained psychological risk. This highlights the necessity for schools to identify such anxiety early and implement systematic interventions to address it.