Cultivating Agency and Responsible AI Use Through the AI-ARC Framework

Synthesis: Wood and Moss propose the AI-ARC framework (Ask, Reflect, Create) as a simple, student-facing alternative to educator-centric AI guidance. Their argument is normative, not empirical: responsible AI use is not a limit on Learner Agency but its strongest expression, and students act ethically by using AI with judgment, reflection, and purpose rather than avoiding it. AI-ARC adapts the authors' earlier AI-ICE model — itself derived from the ICE model of Ideas, Connections, and Extensions — shifting its purpose from evaluating engagement with AI to guiding it. Ask replaces Ideas, Reflect replaces Connections, and Create replaces Extensions, moving Learners from curiosity through critical and ethical evaluation to original authorship. Drawing on Bandura's account of agency, the paper argues that AI efficiency can erode Critical Thinking and offers designers a heuristic: map objectives to ARC stages, build reflection checkpoints, and treat responsible use as part of academic integrity. The framework is proposed, not tested.

Key Findings

• AI-ARC stands for Ask, Reflect, Create, a three-stage student-facing scaffold adapted from the authors' earlier AI-ICE model and the ICE model of Ideas, Connections, and Extensions.
• Ask replaces Ideas, emphasizing curiosity and intentional questioning so students surface multiple possibilities from AI rather than passively accepting the first output.
• Reflect replaces Connections, asking students to judge outputs against course concepts, disciplinary sources, and ethics, including bias, accuracy, and authorship.
• Create replaces Extensions, the synthesis stage where learners transform AI-assisted material into original work, asserting authorship and integrating personal voice.
• Drawing on Bandura (1999), the authors define student agency as purposeful choices about how, when, and why to use AI — monitoring thinking, questioning outputs, owning products.
• Responsible AI use is ethical, transparent, accountable practice preserving human judgment; the authors stress it is reflective self-regulation rather than rule-following.
• Prescribed practices include step-by-step prompts, ethical checklists, reflection journals documenting AI use, and peer critique of AI-assisted outputs.

Why the goal is agency rather than avoidance

The paper's framing move is to reject equating responsibility with restriction. Wood and Moss argue that environments where AI assumes too much of the cognitive load may limit students' ability to engage deeply in Problem Solving, and that favoring convenience over reflection lets AI take on cognitive effort that would otherwise belong to the learner. Their answer is not to withhold the tool but to design tasks requiring active evaluation of AI outputs, challenge to assumptions, and human judgment applied to improving results. Responsibility is thus the exercise of autonomy, not its surrender. They distinguish this from pseudo-agency, citing González-Howard et al. (2024), where students appear engaged and follow procedures without understanding the epistemic purpose behind their actions — in AI collaboration, using outputs uncritically without questioning their validity.

What each AI-ARC stage asks of students

Ask is exploratory: students use AI to surface possibilities, generating several approaches to a topic, examples to compare or critique, and perspectives that may be missing. Agency here is choice and curiosity about what to ask. Reflect is evaluative: students ask how an output aligns with what they have learned, what assumptions or biases it carries, and which parts are supported by evidence versus needing validation. Agency here is judging information and connecting it to prior knowledge and ethical standards. Create is authorial: students ask how to revise material into their own argument and what original insight to add. Agency here is asserting authorship and owning the final contribution. The authors describe the arc as a trajectory that begins with curiosity, deepens through ethical reflection, and culminates in creative action.

Designing instruction around the arc

For instructional designers, the framework becomes a planning routine. During analysis and design, designers map course objectives to each ARC stage to determine where AI should support inquiry, where human judgment must remain central, and where students synthesize independently. During development, the stages guide prompts, reflection checkpoints, and assignment scaffolds; during evaluation, they become criteria for reviewing whether a course supports ethical AI use and preserves student authorship. In the classroom, the authors recommend Scaffolding AI use without undermining autonomy — step-by-step prompts, an ethics checklist, or reflection journals — with students free to decide how deeply to engage at each stage. Peer collaboration is suggested so students compare and critique AI-assisted outputs together, and the authors urge aligning the framework with academic integrity policy rather than treating responsible use as a separate compliance burden.

What this means for practice

• Treat AI policy as a design problem: decide where AI supports inquiry, where human judgment stays central, and where students must synthesize alone.
• Build reflection into assignments so students document how they used AI and how they ensured authorship in the final product.
• Scaffold without rescinding autonomy — provide prompts, checklists, and journals, but let students choose engagement depth.
• Use the ARC stages as criteria when reviewing whether a course supports ethical AI use and preserves student authorship.

Limitations

• The paper is conceptual; the authors report no study, participants, or outcome data testing whether AI-ARC changes student behavior or learning.
• The framework is adapted from the authors' own earlier AI-ICE model, so its refinement is argued by its originators rather than validated independently.
• The classroom and disciplinary examples (writing, biology, teacher education, business) are illustrative suggestions, not documented implementations with results.

Connected Concepts

• Learner Agency
• Critical Thinking
• AI Literacy
• Self-Regulated Learning
• Metacognition
• AI Use and Disclosure Statements

Connected Articles

• A Student-Centered Framework for Responsible Use of Generative AI in Higher Education — another student-facing framework for responsible GenAI use in learning.
• The LEARN Framework for Responsible Use of Generative AI in Education: A Neuroscience-Informed Model for Problem-Based Learning — a responsible-GenAI scaffold applied through project-based learning.
• From Mechanical Compliance to Human Flourishing: A Socialist Humanist Approach to Asynchronous AI Literacy and Fair Use in Higher Education — challenges compliance-driven AI literacy in favor of human flourishing.
• Students' Agency in GenAI-Mediated Group Assessment: An Ecological-Emergent Perspective — examines learner agency and generative AI in group assessment settings.

Citation

Wood, D., & Moss, S. (2026). Cultivating Agency and Responsible AI Use Through the AI-ARC Framework. Journal of Instructional Design and Technology, 1(1), 8-17.