Meta-analysis reveals generative AI tools enhance self-efficacy in learners, indicating that interactive tutoring roles yield the greatest motivational benefits.
Key Points
To evaluate the impact of generative artificial intelligence tools on learners' self-efficacy across experimental and quasi-experimental studies.
Systematically searched and screened experimental and quasi-experimental studies published between November 2022 and February 2026.
Meta-analyzed 84 independent effect sizes examining generative AI use across varying intervention durations, instructional roles, academic domains, and age groups.
Generative AI use during learning produced a moderate positive effect on learner self-efficacy.
Interventions that positioned generative AI as an instructional tutor guiding students generated larger self-efficacy gains than those limited to feedback, answering questions, or replacing student work.
Longer interventions yielded stronger self-efficacy improvements than brief interactions, with benefits remaining consistent across academic subjects and age cohorts.