Scaffold or Shortcut? Postgraduate IT Students' Use of Generative AI and Self-Regulated Learning

Synthesis: Scaffold or shortcut? — a mixed-methods study of 267 postgraduate IT students in Australia showing that GenAI plays a dual role in SRL: it can act as a cognitive scaffold that supports goal setting, monitoring, and self-evaluation, or a cognitive shortcut that displaces them. Whether it scaffolds or short-circuits depends less on the tool than on how students regulate its use.

Key Findings

• GenAI's dual role. Students used GenAI both as a practical aid and a cognitive partner — supporting goal setting, reflection, and feedback, yet at times displacing monitoring and self-evaluation when used uncritically.
• Scaffolded vs. substitutional Cognitive Offloading. The study distinguishes two modes: scaffolded offloading (students use GenAI to clarify goals, generate ideas, or obtain feedback that they then critique and adapt — agency stays with the learner) versus substitutional offloading (students accept outputs with minimal verification, shifting control to the tool).
• Patterns were differentiated, not homogeneous. Most students gained confidence and AI literacy over the term, but some became confident, regular users while others stayed hesitant; postgraduate cohorts are not a homogeneous "AI-literate" group.
• Confidence shaped orientation. Confident users demonstrated agency and autonomy in reflective goal setting and monitoring; less confident students were more likely to view GenAI as a shortcut or form of academic misconduct.

What this means for practice

• Instructors. Require students to justify or adapt AI outputs in the submitted work, so that scaffolded offloading — clarifying goals, generating ideas, critiquing feedback — is visible and substitutional offloading can be caught.
• Instructors. Teach AI literacy as critical, strategic, and responsible engagement rather than tool operation, because the students who treated GenAI as a shortcut or as misconduct risk were the less confident users.
• Learners. Keep Learner Agency over the SRL cycle: use GenAI to support goal setting, monitoring, and self-evaluation rather than handing those phases to the tool.
• Faculty developers. Plan for differentiated cohorts instead of a homogeneous "AI-literate" intake, since confidence and use patterns diverged across the term even as most students gained confidence.

Limitations

• 267 students in a single postgraduate IT subject at one large Australian university; the authors state the findings are limited to one subject.
• The evidence rests on self-report: three custom surveys across a 12-week collection window (Week 1, Week 9, and a Week 10 reflection) plus interview data from two students, so the scaffold-versus-shortcut distinction reflects reported orientations rather than observed AI behavior.
• The 12-week span captures one semester only, so the movement from hesitant to confident regular user cannot be separated from ordinary course progression.
• The study is a single-subject mixed-methods design with no comparison condition, so the differentiated profiles are descriptive rather than evidence of an instructional effect.

Citation

Atif, A., & Dickson-Deane, C. (2026). Scaffold or Shortcut? Postgraduate IT Students' Use of Generative AI and Self-Regulated Learning. Learning Letters, 8, Article 63.