Mathematical Modelling of Ethical AI Use in Higher Education: A Coordination Game Framework for Future-Facing Learning

A new preprint reframes student use of generative AI in assessments as a coordination problem, arguing that peer expectations and assessment design—not just individual compliance—determine whether cohorts adopt responsible or opportunistic norms [1][2]. The paper, submitted to arXiv on 23 April 2026, develops an evolutionary game-theoretic framework that models how collective AI-use practices emerge and stabilize in higher education [1][2]. The authors capture learning value, effort, perceived fairness, and transparency, while institutional governance is represented implicitly through what they term reflective assessment incentives [2]. Generative AI—systems capable of producing text, images, and other media from prompts—has become widely available since the early 2020s, accelerating an AI boom that has prompted discussions around regulation and ethical use [3]. In education, the technology intersects with a long history of educational technology, or edtech, which draws on disciplines including psychology, sociology, and computer science [4]. The rapid classroom uptake has intensified concerns about academic integrity, fairness, and whether students are actually learning [1][2]. The preprint’s simulations reveal threshold-driven behavioral transitions. Small, well-calibrated changes in reflective assessment incentives can trigger rapid shifts toward responsible, learning-oriented AI-use norms, while weak or misaligned incentives allow opportunistic practices to persist [2]. These non-linear dynamics, the authors contend, explain why institutional policy statements alone often fail to change behavior, whereas modest assessment redesigns can have disproportionate effects [1][2]. The framework offers an analytically grounded tool for what the paper calls Future Facing Learning, supporting pedagogy-led AI governance without reliance on surveillance or punitive enforcement [2]. The work arrives as higher education institutions continue to grapple with how neurodivergent students, among others, experience assessment environments differently—a consideration that intersects with broader debates about inclusive design and the social model of disability [5]. The preprint does not address neurodiversity directly, but its emphasis on assessment structure rather than individual compliance echoes wider calls to reduce person-environment mismatches in educational settings [5].