The Impact of AI on Learners' Self-Efficacy: Insights for Clinical Practice

A meta-analysis of 23 studies found a medium, statistically significant aggregate effect of AI interventions on learner self-efficacy (Hedges' g = 0.758), supporting the use of self-efficacy as a proximal outcome. Direct evaluation in clinical mental-health populations is still needed.

These pooled educational findings endorse self-efficacy as a measurable psychoeducational endpoint, but whether similar effect sizes appear in clinical assessment or psychoeducation for mental-health patients requires trials in those populations.

Interpretation requires caution: heterogeneity across studies was high, signaling substantial between-study variability. Moderator analyses partially accounted for this heterogeneity, so effect magnitude may differ by setting and implementation.

Where were effects largest? Moderator analyses identified substantially higher effects in natural sciences and medicine, whereas engineering produced a non-significant outcome. This pattern suggests AI-driven interventions have greatest impact in practice-oriented, simulation-ready domains.

Effect sizes also varied by coded AI role. Categories labeled 'intelligent learning tool' and 'intelligent tutor' showed larger aggregated effects; however, avoid overinterpreting a more granular taxonomy of roles unless primary studies explicitly support it.

Lastly, application type mattered as simulations and personalized tutoring yielded larger effects than passive content or one-size-fits-all materials.

For clinical psychoeducation, prioritizing interactive, mastery-focused AI—practice opportunities, corrective feedback, and graded challenges—is likely to maximize self-efficacy gains.

Key Takeaways:

• AI produces medium-to-large self-efficacy gains (Hedges' g = 0.758), indicating measurable psychoeducational benefit within educational contexts.
• Largest gains occurred in natural sciences and medicine; engineering showed minimal change.
• Align AI roles (tutoring, simulations, adaptive feedback) with psychoeducation goals—practice, corrective feedback, graded challenges—for greater impact.