GATs produced no reliable immediate learning-experience effects on frustration, cognitive load, or situational interest at either institution¶

Claim · All claims
Evidence · 1 study · causal r2 · q3
Currency · no dated study recorded · checked against its sources 2026-10-16 (automated, 1 of 1 entry) · not reviewed by a person

Subclaims¶

q3 i? Pooled effects on frustration, cognitive load, and situational interest were small and did not survive FDR correction across both deployments. → experience-null

Evidence¶

experience-null¶

Yuri Noviello, Naaz Sibia, Anastasiia Birillo, Thomas Overklift Vaupel Klein, Michael Liut, and Gosia Migut. 2026. AI-Generated Traces for Novice Programmers: Learning Effects and Learner Differences in a Multi-Institutional Study. Proceedings of the 31st ACM Conference on Innovation and Technology in Computer Science Education V. 1 (ITiCSE 2026). https://doi.org/10.1145/3803400.3809346

q3 · i? · causal · r2

Likert self-report instruments (Paas mental effort, NASA-TLX, IMI items) analyzed with linear mixed models at both sites; e.g., UofT pooled frustration Δ=-0.135, p=0.043, padj=0.086, and TU Delft pooled situational interest Δ=0.240, p=0.079, padj=0.196. Equivalence was not tested.

"Across both deployments, pooled effects on frustration, cognitive load, and situational interest were small and did not survive correction."

Discussion¶

Learner Variables¶

• Affect Regulation — outcome: instruction changes it
• Motivation — outcome: instruction changes it
• Interest — outcome: instruction changes it

Related Claims¶

• Engagement profiles significantly moderated GATs' immediate performance effects: low- and high-engagement profiles benefited while the mid-engagement profile showed a small decrement — related
• At UofT, GATs produced a reliable end-of-course increase in Constructive engagement, with no effects on Passive, Active engagement, or MSLQ elaboration — related
• In the Java-based CS1 course at TU Delft, GATs significantly improved immediate post-intervention performance pooled across topics, driven mainly by the While-loop intervention — related
• In the Python-based CS1 course at UofT, GATs produced no significant overall or topic-specific immediate performance effects — related
• Across both institutions, GATs showed no significant effect on long-term summative exam performance — related
• Changes in situational interest were not statistically significant for either Mathbot or BAU students during the five-day fraction intervention — related
• Personalized paths reduced self-reported cognitive load on all six adapted NASA-TLX dimensions with large effect sizes — related
• AI assistance reduces subjective mental effort across all tasks even when it does not reduce completion time, dissociating time and effort — related
• Self-reported cognitive load showed no significant differences across either volume-based or content-based offloading profiles — related
• Related decorative pictures do not increase triggered or maintained situational interest; in Experiment 2 they reduced maintained interest — related