Generative AI (GenAI) as a mindtool that supports generative learning (GL)

Synthesis: Generative AI (GenAI) as a mindtool that supports generative learning (GL)

Key Findings

The Eight Pedagogical Roles

The roles span the learning process. As a learning strategy or study buddy, GenAI supports knowledge organization and comprehension monitoring at varying degrees of complexity; as a collaborative thinking tool, it fosters teamwork and project-based activities by encouraging the sharing, discussion, and integration of spatial representations of content into a more cohesive knowledge structure; as a possibility engine, it generates alternative responses that let students explore different ways of expressing ideas; and as a Socratic opponent, it challenges students to develop and refine their arguments. The remaining roles extend beyond dialogue: a personal tutor provides personalized feedback, an exploratory research engine helps students explore and interpret data, a motivator proposes games and challenges to engage learners, and a dynamic assessor evaluates student knowledge in real time, allowing for tailored generative learning activities based on current understanding.

Evidence from Research

The authors ground the mindtool framing in three strands of evidence. First, the eight generative learning strategies formalized by Fiorella and Mayer (2015) — summarizing, mapping, drawing, imagining, self-testing, self-explaining, teaching, and enacting — operationalize Wittrock's (1974) original generative learning theory, and each maps onto roles GenAI can scaffold. Second, Makransky et al.'s (2025) experiments with a theory-informed chatbot (ChatTutor) that scaffolds student-produced explanations show the value of explicit generative scaffolding over raw LLM use: in Study 1 (N = 175; ChatTutor = 51, ChatGPT-4 = 49, teaching-as-usual = 75), immediate post-test differences were non-significant, F(2,172) = 1.89, p = .155, but four weeks later ChatTutor users retained substantially more conceptual knowledge (M = 7.26 of 9) than ChatGPT users (M = 6.72) or teaching-as-usual (M = 5.93). Third, concept mapping — a primary generative strategy — shows large benefits versus passive learning (college students d ≈ 0.72; grades 4–8 d ≈ 0.68; grades 9–12 d ≈ 0.74), and can now be co-constructed iteratively with GenAI-assisted tools. The authors also note that generative strategies are not universally effective: their fit with learners' age and cognitive capacity matters, and strategic scaffolding — such as partially completed concept maps or metacognitive prompts — makes advanced strategies accessible to all learners.

The A2-GLD Framework

To put these ideas into practice, the paper proposes the AI-Augmented Generative Learning Design model (A2-GLD), a five-phase pedagogical framework that combines GL theory, GenAI affordances, and instructional scaffolding:

1. Prime the learning task — teacher-initiated; activates prior knowledge and curiosity through prompting questions, AI-generated visuals, and analogies (e.g., "What do you already know about ecosystems?").

2. Enact or Visualize — learner-initiated; builds internal mental models through non-verbal generative activities such as concept maps, diagrams, and gesture-based explanations, optionally animated via AI tools.

3. Explain with AI — learner-driven; students verbalize understanding in their own words with AI feedback (e.g., "Great start! Can you clarify what you mean by energy transfer?").

4. Human Scaffolding — instructor-driven; teachers review AI–student interaction artifacts, facilitate peer discussion, and embed metacognitive prompts (e.g., "What confused you the most in the AI chat?").

5. Reflect and Transfer — learner-driven; strengthens metacognition and transfer of learning beyond the activity.

The alternating driver of each phase (teacher → learner → learner → teacher → learner) is deliberate: it keeps the instructor in the loop while preserving the learner's active generation of knowledge, and it positions GenAI as a configurable design material — a "primed" tool customized for each activity — rather than a fixed answer machine.

Implications for AI in Education

The framework gives Instructional Design practitioners a vocabulary for using GenAI to enhance rather than replace learning: each role is tied to a specific generative learning function, and the resulting pedagogical model is intended to guide the design of GLAs. This connects to Constructivist and Self Regulated Learning traditions, where the learner's active knowledge construction — supported, not performed, by the tool — is the point of the activity, and it offers a counterweight to answer-generating uses of GenAI in the classroom. The authors frame the underlying goal in terms of learning with, not from, the technology: using GenAI as a mindtool can sharpen inference-making and critical thinking while avoiding the accumulation of inert (unusable) knowledge, and the challenge is to protect learners' creativity and reasoning rather than outsource them. The A2-GLD phases give faculty a concrete route to "design for AI" — preparing customized GenAI tools for specific generative learning activities, with the instructor explicitly present in the Human Scaffolding phase.

Connected Concepts

Connected Articles

Citation

Dabbagh, N., & Fake, H. (2026). Generative AI (GenAI) as a mindtool that supports generative learning (GL).