Impact of AI assistance on student agency
Adults / lifelong learning AI and learning Feedback and assessment
Links to claims
Supports (1)
| Claim | Basis |
|---|---|
| C2 | Indirect: RCT with 1,625 university students: after four weeks of AI prompts on peer-feedback quality, students relied on rather than learned from the AI; quality dropped without it and checklists only partly filled the gap (abstract). Skill outcome, not a closed-book knowledge test; full text not read. |
Review state: first-pass draft, pending human review.
Source note
Syntes: Darvishi et al. (2024) — students relied on AI prompts rather than learning from them
Källa: Darvishi, A.; Khosravi, H.; Sadiq, S.; Gašević, D.; Siemens, G. (2024 (Computers & Education 210, 104967)). Impact of AI assistance on student agency. https://doi.org/10.1016/j.compedu.2023.104967
What the abstract says (read 2026-10-11; full text not read)
- Design: randomised controlled experiment, 1,625 students across 10 courses, peer review. First four weeks: AI features (rule-based suggestion detection, semantic similarity, comparison with the reviewer's previous comments) prompted students to improve feedback that was too general. Next four weeks, four groups: AI prompts (control), no prompts (NR), self-monitoring checklists instead (SR), AI prompts plus checklists (SAI).
- Result (abstract wording): "students tended to rely on rather than learn from AI assistance". When AI assistance was removed, self-regulation checklists helped fill the gap but were not as effective as AI assistance; AI plus checklists was not more effective than AI alone.
- Not verified here: effect sizes and exact measures; read the full text before using in a verdict.
Why it is in the library
Adjacent (human–AI interaction, student agency): a large randomised study of what happens when AI scaffolding is withdrawn. It is the same pattern as C2 (performance with the tool is not learning) in a skill (writing useful feedback) rather than a knowledge test.
Metadata as recorded
- ID
- AI-026
- Tier
- RCT
- Closed-book follow-up
- n/a (outcome is the quality of the students' own peer feedback once AI prompts are removed, not a knowledge test)
- Titel
- Impact of AI assistance on student agency
- Författare
- Darvishi, A.; Khosravi, H.; Sadiq, S.; Gašević, D.; Siemens, G.
- År
- 2024 (Computers & Education 210, 104967)
- Källa/DOI
- https://doi.org/10.1016/j.compedu.2023.104967
- URL
- https://doi.org/10.1016/j.compedu.2023.104967 (open access, CC BY)
- Typ
- randomised experiment with 1,625 university students in 10 courses writing peer reviews on a learning platform; four weeks with AI prompts on feedback quality for everyone, then four weeks in four groups: AI prompts continued, no prompts, self-monitoring checklists instead, or both
- Åldersgrupp
- vuxen-livslangt
- Teman
- ai-och-larande;feedback-bedomning
- Hämtad
- 2026-10-11
- Paywall
- open access (CC BY), not reachable from the box
- Tillagd
- source sweep 2026-10-11 (kind: adjacent); links pending human review
- Full text (fetched 2026-10-16, legal open-access copy, private)
- fulltext.pdf + — via unpaywall: https://researchmgt.monash.edu/ws/files/572678646/564026501_oa.pdf
The original files (PDF or abstract) are kept in the project archive and are not published here.