The LEARN framework for responsible use of generative AI in education: a neuroscience-informed model for problem-based learning

Abstract In an era of rapid generative artificial intelligence (GAI) integration into education, students are increasingly using these tools not merely as learning aids but as their primary means for completing assessments. This shift raises significant concerns regarding academic integrity, cognitive offloading, and the erosion of critical thinking. To address these challenges, this paper advances a conceptual, neuroscience-informed framework, the Lifelong Learning, Engagement, Active Processing, Reflection, and Neuro-based Design (LEARN) model, for the ethical and pedagogically grounded integration of GAI into assessment contexts. Grounded in over two decades of experience with problem-based learning (PBL), the framework emphasises learner autonomy, adaptability, and sustained cognitive engagement. The LEARN framework synthesises principles from cognitive and educational neuroscience with constructivist learning theory to explain how learning processes such as neuroplasticity, effortful cognition, metacognitive regulation, and socio-emotional engagement can be intentionally supported in AI-mediated environments. Each component positions GAI as a cognitive scaffold rather than as a cognitive substitute, encouraging critical evaluation, reflective judgement, and ethical self-regulation. By integrating neuroscience-informed learning design, PBL pedagogy, and responsible AI use, the LEARN framework contributes a theoretically grounded model for redesigning assessment practices that sustain deep, self-directed, and reflective learning in the context of generative AI.