Shallow, copy-pasting engagement occurred but was not the dominant mode of GenAI Tutor use, and was lowest during exam preparation¶

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

Subclaims¶

q2 i? Shallow engagement accounted for 10.4% of sessions, and exam-driven sessions exhibited the least copy-pasting of the four types. → Youjie Chen 2026

Evidence¶

Youjie Chen 2026¶

Youjie Chen, Xixi Shi, Xinyu Liu, Shuaiguo Wang, Tracy Xiao Liu, Dragan Gašević. (2026). Not All Students Engage Alike: Multi-Institution Patterns in GenAI Tutor Use. arXiv preprint. https://arxiv.org/abs/2602.00447 doi:10.48550/arxiv.2602.00447

q2 · i? · associational · r2

Authors' interpretation of the session-level clustering results, in which shallow engagement was 10.4% of the 113,255 sessions and exam-driven engagement (31.2% of sessions) "exhibited the least copy-pasting behaviors." The authors conclude students often seek broad learning support rather than simple answer extraction.

"We show that shallow engagement, as commonly characterized by copy-pasting behaviors and direct answer requests, did occur but was not the dominant mode of GenAI Tutor use."

Discussion¶

Learner Variables¶

• Motivation — outcome: instruction changes it

Related Claims¶

• Four distinct session-level engagement types emerge in GenAI Tutor use: Deep, Shallow, Routine-Learning, and Exam-Driven — related
• Students at highly selective institutions show more deep and routine-learning engagement and less exam-driven engagement with a GenAI Tutor — related
• STEM students adopt the GenAI Tutor less and use it differently than non-STEM students, with more shallow and routine-learning engagement — related
• Shallow engagement with a GenAI Tutor persists over time, while deep engagers transition flexibly across engagement types — related
• Student chatbot queries were overwhelmingly task-oriented initial inquiries, dominated by definition requests and pasted assignment questions rather than iterative critique — reports the opposite
• Most unrestricted-mode interactions were direct answer retrieval: 52.9% of exercise interactions were copy-paste of the question, and participant answers were linguistically near-identical to chatbot text in 25.3% of cases — reports the opposite
• GenAI's default output on a take-home exam question remained at the quantitative, multistructural SOLO phase, producing polished but pedagogically thin responses — related
• Students' GenAI prompts overwhelmingly exact-copy instructions and request information rather than review — reports the opposite
• Guest-mode usage dominates lightweight help-seeking, raising an equity concern that students without metacognitive support stay at shallow answer lookup — related