AI access shifts learning-time composition away from writing toward reading and searching, with total learning time unchanged¶

Claim · All claims
Evidence · 1 study · causal r3 · q4
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q4 i? Treated students spend 5.3 pp less of their learning time on writing activities and 4.4 pp more on research activities, while total time spent learning does not change significantly. → Zara Contractor and Germán Reyes 2026

Evidence¶

Zara Contractor and Germán Reyes 2026¶

Zara Contractor and Germán Reyes. (2026). Experimental Evidence on the Learning Impact of Generative AI. https://arxiv.org/abs/2607.08849 doi:10.48550/arxiv.2607.08849

q4 · i? · causal · r3

Self-reported time-allocation shares from the post-learning survey in the randomized experiment; Qualtrics-recorded learning-phase duration changes by only 0.2 minutes and self-reported duration by 0.5 minutes, neither significant. The article reports corresponding TOT effects of a 7.7 pp writing-share decrease and a 6.4 pp research-share increase.

"Treated students spend about 5.3 pp less of their learning time on writing activities (drafting, editing, note-taking, and organizing), down from a control-group share of 53.4 percent (p= 0.028), and about 4.4 pp more on research activities (reading and searching), up from 44.5 percent (p= 0.050)."

Discussion¶

Related Claims¶

• Practicing with AI was less effortful than practicing alone, with participants spending less time and logging fewer keystrokes — related
• Revisions made no reliable difference to total time spent with the lesson — related
• Random assignment to AI access produces a strong first stage: about 70 percent of treated students use AI, most often to explain concepts — related
• Engage-to-Unlock redistributed effort across tasks: participants spent more time writing and less time evaluating, without increasing overall task duration — related
• AI-tutored students spend less time on task than the 60-minute in-class lesson, with no correlation between time and scores — related
• Guided LiquidText users redistribute task time toward document processing and away from writing — related
• Total learning time did not differ, but the worked-example group tended to spend more time on the most relevant quality-criteria content — related