Beyond Task Completion: A Theoretical Integration and Framework for Guiding Students' ChatGPT Use for Learning

Synthesis: Øby (2026) develops an explanatory framework for students' ChatGPT use grounded in learning theory, integrating cognitive load theory and goal orientation (Motivation) with Self-Efficacy and task relevance as mediating constructs. The core claim is that generative AI can either support learning or enable superficial task completion depending on how it is integrated into academic work. Four perspectives — cognitive demands, motivational orientations, perceived competence, and perceived task value — shape students' decisions about when and how to rely on genAI, determining whether it functions as a scaffold supporting thinking and knowledge construction or a shortcut that bypasses essential cognitive processes.

Key Findings

• Scaffold versus shortcut: Whether generative AI supports learning or enables superficial task completion is contingent on cognitive and motivational conditions shaped within instructional contexts — not inherent to the tool.
• Cognitive load as the mechanism: ChatGPT can reduce extraneous load by structuring information and activating prior knowledge, but when outputs replace engagement with interacting elements, intrinsic load is bypassed rather than managed. Task complexity is closely tied to element interactivity.
• Motivational orientations drive use: Mastery and normative performance goals tend to support AI use for elaboration and refinement, whereas appearance performance goals increase the likelihood AI is used to minimize effort and protect perceived competence — a channeling of achievement goals through engagement with academic tasks.
• Self-Efficacy mediates: Students who believe they can succeed in demanding tasks are more likely to sustain engagement and adopt mastery-oriented strategies; attributing success to external tools rather than effort constrains efficacy growth and reinforces shortcut patterns.
• Task relevance shapes investment: When tasks are perceived as central to academic or professional development, students tolerate intrinsic load and use AI to elaborate understanding; when peripheral, efficiency concerns dominate and AI is used to reduce effort.
• AI literacy as Pedagogies and Teaching Strategies: The framework positions genAI use as an instructional variable, suggesting that educators can design learning conditions that manage cognitive load and support adaptive orientations rather than relying on reactive policy restrictions.

What this means for practice

• Instructors. Sequence academic tasks with gradually increasing element interactivity so intrinsic load stays aligned with learner expertise, because ChatGPT's value flips between scaffold and shortcut depending on task complexity.
• Instructors. Frame generative AI explicitly in course design and classroom communication, discussing when it supports reasoning and when it replaces it, instead of leaving guidance to reactive Academic Integrity policy.
• Instructors. Break complex activities into incremental steps so students experience manageable successes and attribute progress to their own strategies, strengthening Self-Efficacy and sustaining engagement with core reasoning.
• Designers. Treat ChatGPT as a cognitive tool for exploring ideas and refining reasoning — structured much as established metacognitive supports are — and shape the motivational climate, since achievement goals respond to contextual cues.
• Faculty developers. Position AI literacy as pedagogical knowledge grounded in learning theory, so instructors can design conditions that manage cognitive load rather than only policing AI use.

Limitations

• This is a conceptual, theory-driven synthesis, and the author states that the proposed relationships among cognitive load, goal orientation, self-efficacy, and task relevance require systematic empirical investigation; the framework is not an empirically validated intervention.
• No sample, intervention, or outcome data are reported, so the design can support hypothesis generation and instructional design but not causal claims or effect sizes for the scaffold-versus-shortcut pathway.
• No measurement instrument or validation is specified for the mediating constructs, so the framework cannot yet be used to measure whether a given course pushed students toward scaffolding or task completion.
• The author reports using ChatGPT for grammar, text flow, and APA alignment and Perplexity AI to assess the framework, making part of the synthesis itself AI-assisted.

Citation

Øby, E. (2026). Beyond task completion: A theoretical integration and framework for guiding students' ChatGPT use for learning . Journal of University Teaching and Learning Practice, 23(5).