Most students reported no change in study habits, but students who adapted their study strategies rated GenAI-evaluation assignments as significantly more helpful¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r2 · q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? 77.0% of surveyed students (154/200) reported the modified homework did not change how they studied for exams. → Ethan Dickey 2026 q3 i? Helpfulness ratings differed significantly across study-change categories (Kruskal-Wallis H(2)=17.74, p<.001), with adapters rating grading more helpful than non-changers (both Bonferroni-adjusted pairwise comparisons significant). → Ethan Dickey 2026 (2)

Evidence¶

Ethan Dickey 2026¶

Ethan Dickey, Marios Mertzanidis, Alexandros Psomas. (2026). Is Solving Better Than Evaluating GenAI Solutions? https://arxiv.org/abs/2607.27586 doi:10.48550/arxiv.2607.27586

q3 · i? · causal · r2

Post-final survey (N=200, 91% response rate) with a three-point item on whether the modified homework changed exam preparation; helpfulness showed a neutral median rating (3.0) in both groups.

"the majority of students reported no change (154/200, 77.0%), followed by "Kind of" (32/200, 16.0%) and "Yes" (14/200, 7.0%)."

Ethan Dickey 2026 (2)¶

Ethan Dickey, Marios Mertzanidis, Alexandros Psomas. (2026). Is Solving Better Than Evaluating GenAI Solutions? https://arxiv.org/abs/2607.27586

q3 · i? · causal · r2

Association analysis on the post-final survey; pairwise Mann-Whitney tests with Bonferroni adjustment showed No vs. Kind of (padj=.019) and No vs. Yes (padj=.002) significant, Kind of vs. Yes not significant (padj=.122).

"A Kruskal–Wallis test indicated a statistically significant difference in helpfulness ratings across the three study-change categories (𝐻(2)=17.74,𝑝<.001)."

Discussion¶

Related Claims¶

• Students received significantly higher homework scores during GenAI-evaluation periods, but this advantage did not transfer to downstream exams — related
• Perceived usefulness of AI and intent to continue using it differ significantly across challenge domains, with academic tasks rated most useful — related
• Design students report low trust in GenAI outputs and systematically modify and verify them, with inaccuracy the main deterrent — related
• GenAI-condition students rated their PBL tutorial significantly higher after the tutorial than they rated their previous ePBLM-based experiences, especially clinical accuracy — related
• Midterm self-efficacy (comfort and confidence) did not differ between conditions after three assignments, though both varied significantly across algorithmic topics — related