Research Summary: The regulation paradox: agentic AI, bounded autonomy, and self-regulated learning in higher education

The emergence of agentic AI in educational technology, moving beyond reactive assistance to systems capable of planning and adaptation, presents a significant paradox for higher education. This development prompts a critical question regarding whether increased AI autonomy fosters or hinders students' self-regulated learning (SRL) capabilities. The analysis identifies that agentic AI is pedagogically valuable when its autonomy supports the learning process itself, rather than solely focusing on academic outcomes. It also highlights four distinct risks: regulatory displacement, metacognitive laziness, reduced productive struggle, and epistemic over-trust, which require careful consideration.

Why it matters

The integration of agentic AI into educational frameworks carries profound implications for the development of learner autonomy and critical thinking skills. Effectively managing the 'regulation paradox' is crucial for ensuring that technological advancements enhance human capabilities rather than diminishing them, thereby impacting long-term educational outcomes and workforce preparedness.

Key insights

• Agentic AI transforms educational technology by introducing capabilities for planning, memory, adaptation, and proactive initiation of actions.
• A core pedagogical question arises: does greater AI autonomy enhance self-regulated learning (SRL) or usurp essential regulatory practice from students?
• The educational value of agentic AI is realized when its autonomy supports learning processes, not merely academic output.
• Four specific risks associated with the 'regulation paradox' are identified: regulatory displacement, metacognitive laziness, reduced productive struggle, and epistemic over-trust.

Source

Frontiers in Education — https://www.frontiersin.org/articles/10.3389/feduc.2026.1902833