As universities rush to roll out generative AI policies and train staff in prompt writing, a new paper in Frontiers of Digital Education warns that faculty development is too often focused on tool demonstrations, compliance advice and risk management—leaving the more consequential question unaddressed: Whether AI will amplify student thinking or displace it.
The conceptual review, by Tendai Charles and Marwa Elkhouly of The British University in Dubai, proposes the PEDAL model—purpose, epistemic challenge, dialogue, accountability and learning evidence—as a practical framework for faculty development in AI-rich higher education. Rather than asking whether AI can complete academic tasks (it increasingly can), the authors argue the real question is how teaching can be designed so that AI supports—rather than short-circuits—analysis, evaluation, creation, metacognition and transfer.
“GenAI is less a wholly new disruption of learning than a stress test of pedagogy,” the authors write. It “reveals weaknesses in constructive alignment, task authenticity, assessment validity, and the professional support available to faculty.”
PEDAL positions AI as a bounded pedagogical resource: a generator of alternatives, simulations, misconceptions and prompts for critique—not a replacement for disciplinary judgment. The five principles are sequenced rather than merely listed. Purpose requires academics to specify the exact intellectual operation and quality criteria before selecting a tool. Epistemic challenge designs tasks—conflicting evidence, local constraints, plausible error—where human judgment remains indispensable. Dialogue structures learner–AI–peer–tutor interaction as iterative inquiry rather than answer retrieval. Accountability covers verification, disclosure, privacy and authorship. Learning evidence asks what proportionate traces of reasoning, revision and transfer demonstrate learning rather than merely polished output.
The paper offers worked redesigns in the social sciences and engineering—for example, turning a conventional policy essay into an evidence-tested judgment task in which AI generates three contrasting stakeholder memos, students verify consequential claims, and a brief in-class transfer item tests whether their evaluative criteria travel beyond the rehearsed case. It also sets out feasibility safeguards on workload, privacy, accessibility and fairness, and distinguishes minimum program-level conditions from more mature institutional support.
The authors are explicit that PEDAL is a “meso-level design heuristic,” not a new general theory of learning, and that it remains empirically untested; the worked examples are design propositions, not reports of implemented interventions or student outcomes. They also stress that purposeful non-use of AI is compatible with the model, and that in contexts of unequal access, lecturer-supplied outputs or non-AI routes can preserve the same learning outcomes.
Faculty development, they conclude, must be treated as design development: “The object of professional learning is not merely the tool or policy; it is the capacity to create and evaluate learning arrangements in which human judgment remains educationally indispensable.”
DOI: 10.1007/s44366-026-0108-9