Data Unification, Marginalised Groups and Diminished Experience in AI-Supported Education

This article extends a prior classroom simulation of cumulative educational disadvantage into the artificial intelligence context. The earlier model showed that students with unstable home lives were left behind by ordinary school events because the events rewarded hidden stability. This paper argues that AI-supported education can reproduce the same structure at the level of data, representation, access and recognition. Large language models and educational AI tools may not deliberately exclude marginalised students, but systems trained and designed around dominant data environments can diminish the experience of groups whose languages, cultures, disabilities, religions, geographies, family obligations and practical knowledge are less represented. The article identifies marginalised student groups likely to experience diminished AI support and proposes classroom and policy safeguards. It argues that AI equity requires more than access to devices; it requires recognition of whose lives, values and knowledge are assumed by the system.