Study learning with AI by varying the interactional arrangement and measuring process and delayed reduced-support outcomes¶

Strategy · All strategies
Evidence · 1 claim (1 mixed) · 1 study (1 theoretical), q1 · 0 of 1 report an effect size · 1 claim rests on one study

Description¶

This strategy operationalizes Agentivism for research design. The article argues that "studies of learning with AI should stop treating \"AI use\" as a single treatment condition", because whether AI gives answers, hints, critiques, or multi-step reasoning alters how agency is distributed and how much verification is required. Research should measure not only immediate task outcomes but also the processes through which learners delegate, monitor, revise, and later perform under reduced support, with delayed explanation, adaptation to novel tasks, and independent performance as central outcomes.

Design Implications¶

Context¶

Requirements¶

• Studies must specify the interactional arrangement through which AI enters the task rather than treating AI use as one binary condition.

Constraints¶

• The article notes much current evidence comes from relatively short tasks or brief interventions, so longitudinal work is especially important.

Target Learners¶

• Researchers studying learning with generative and agentic AI

Target Learning Goals¶

• Distinguishing assisted performance from durable human learning in empirical studies

Affordances¶

• Agentivism Learning Theory

Claims¶

• The article proposes an empirical validation agenda: comparing perceived competence under LLM-assisted versus unaided conditions, with measurement frameworks and longitudinal designs [~W]

Related Strategies¶

• Use inclusive design, calibrated supports, and clear delegation norms to preserve learner agency and epistemic diversity
• Research agenda: use process-oriented assessments, study encoding-consolidation-retrieval effects, and examine longitudinal outcomes of educational generative AI

Examples¶

-

Key Sources¶

• Lixiang Yan and Dragan Gašević. (2026). Agentivism: a learning theory for the age of artificial intelligence. https://arxiv.org/abs/2604.07813