Grade effects of GenAI availability do not differ across terciles of students' prior academic preparation, robust across multiple ability proxies¶

Claim · All claims
Evidence · 1 study · causal r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Estimating the DiD separately within terciles of course-residualized within-cohort first-term GPA rank yields null grade effects in each tercile, robust across alternative ability measures. → Dumlao 2026

Evidence¶

Dumlao 2026¶

Dumlao, J. M. Z., Wang, M., Xie, Z., Hu, J., Bar, I., Chaney, G., III, Gold, H., & Teplitskiy, M. (2026). Generative AI Availability, Grades, and Student Satisfaction at a Large University. arXiv. https://arxiv.org/abs/2607.21534 doi:10.48550/arxiv.2607.21534

q2 · i? · causal · r2

Heterogeneity analysis grouping students into terciles by AbilityRank (course-residualized, within-cohort first-term GPA rank), estimating the preferred specification separately per tercile; the article reports "null grade effects across prior academic performance terciles". Detailed tercile results fall beyond the supplied text.

"We also find null grade effects across prior academic performance terciles, robust across our various proxies."

Discussion¶

Related Claims¶

• ChatGPT exposure effects vary across the school performance distribution, with positive estimates for lower-performing schools and negative for top-performing schools, but this pattern is not robust — related
• GenAI availability produced no significant differential effect on final grades in more GenAI-susceptible courses relative to less susceptible ones — related
• Time-varying educational ChatGPT exposure shows no significant effect on high school test scores in the balanced panel; significant unbalanced-panel results are not robust — related
• More GenAI-susceptible courses show no raised grade floor after ChatGPT: passing-margin and failing/withdrawal probabilities are null under both COVID assumptions — related
• Positive average grade effects appear only under COVID-laden treatment anchors, indicating such readings reflect the COVID shock rather than ChatGPT — related