Evidence for emotional-only AI support is limited to two studies, which were summarized narratively rather than pooled¶

Claim · All claims
Evidence · 1 study · quant-synthesis r2 · q2
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Only two studies of emotional-only AI support met meta-analysis criteria; one showed g = 0.35 on task performance and one g = 1.32 on behavior (time on task), and effects on knowledge acquisition were not examined. → Heng Zhang 2025

Evidence¶

Heng Zhang 2025¶

Heng Zhang, Yuhan Liu, Meilin Jiang, Juanjuan Chen, Minhong Wang, Fred Paas. (2025). Emotional Artificial Intelligence in Education: A Systematic Review and Meta-Analysis. Educational Psychology Review. https://doi.org/10.1007/s10648-025-10086-4

q2 · i? · quant-synthesis · r2

Category 3 (emotional-only support) meta-analysis set: VanLehn et al. reported g = 0.35 on task performance and Lin et al. reported g = 1.32 on learning time via fixational eye movements. The authors "did not pool effects" given only two studies.

"Only two studies met criteria (see Appendix 2). We therefore did not pool effects and instead summarized narratively."

Discussion¶

Related Claims¶

• Cognitive-only AI support yields medium effects on knowledge (g = 0.52), skill (g = 0.44), and affect (g = 0.48) and a large effect on perception (g = 0.85) — related