In both interaction-log and think-aloud records from a GenAI-assisted instructional-design task, regulation was the most frequent operational label and monitoring the least frequent¶

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

Subclaims¶

q2 i? Among retained semantic units, regulation was most frequent and monitoring least frequent in both the interaction-log record (41.6% regulation, 35.4% planning, 23.0% monitoring) and the think-aloud record (39.4%, 35.7%, 24.9%). → Li Shijin & Liu Junjian 2026

Evidence¶

Li Shijin & Liu Junjian 2026¶

Li Shijin & Liu Junjian. (2026). Mapping the Distribution and Depth of Metacognitive Processes in Generative AI-Assisted Learning: Evidence from Interaction Logs and Concurrent Think-Aloud Protocols. Behavioral Sciences. https://doi.org/10.3390/bs16081365

q2 · i? · qualitative · r2

Descriptive analysis of coded records from 79 undergraduates using DeepSeek during a 90-minute instructional-design task: 274 interaction-log and 193 think-aloud retained instances. The article reports "regulation accounted for 41.6% of retained instances" in logs and the listed think-aloud percentages; no inferential test was used because instances are nested within participants.

"In the interaction-log record, regulation accounted for 41.6% of retained instances, planning for 35.4%, and monitoring for 23.0%. In the think-aloud record, the corresponding percentages were 39.4%, 35.7%, and 24.9%."

Discussion¶

Related Claims¶

• Evidence-depth ratings in both records were concentrated at the awareness and strategic labels, with implicit and reflective evidence uncommon — related
• Retained monitoring units showed a higher combined strategic-plus-reflective share in think-aloud speech than in interaction logs (66.7% vs 41.3%) — related
• Retained planning units showed a higher combined strategic-plus-reflective share in think-aloud speech than in interaction logs (59.4% vs 49.5%) — related
• Retained regulation units showed a higher combined strategic-plus-reflective share in interaction logs than in think-aloud speech (69.3% vs 55.3%) — related
• Predicting, using contextual analysis, and rereading were the most frequently used metacognitive behaviors; deciding when to stop, independent reading time, and testing were least frequent — related
• Evidence suggests over-reliance on generative AI may negatively affect students' self-regulated learning behaviors, potentially harming retention and transfer to novel contexts — related