Evaluating the Impact of Generative AI-Driven Adaptive Assessment Frameworks on Critical Thinking and Academic Integrity in Higher Education

Abstract

Generative Artificial Intelligence (GenAI) adaptive assessment, in which the difficulty of questions, hints and feedback adjust to a learner’s live performance, is increasingly presented as a means of supporting personalised learning and critical thinking. At the same time, it may encourage cognitive offloading and create new academic-integrity concerns. This pilot study examines student perceptions of GenAI adaptive assessment, critical thinking, cognitive offloading, academic integrity and institutional readiness. The study combines secondary literature with a primary Google Form survey of 13 higher education students collected on 14 September 2026 through convenience sampling. Responses were analysed using frequencies, descriptive statistics, one-sample Wilcoxon signed-rank tests, sign tests, rank correlations and thematic coding. Respondents agreed that interactive AI hints improve the ability to break down multi-step problems (69.2%; p = .011) and that excessive reliance on real-time AI causes cognitive offloading (61.5%; p = .005). They also considered cognitive offloading in unmonitored work frequent (61.5%; p = .010). However, respondents were undecided on whether process-based adaptive assessment reduces ghostwriting, and they were divided on the clarity of institutional guidelines. As the sample is small, non-random and self-reported, the findings are indicative rather than generalisable. The study recommends adaptive systems designed around hints rather than final answers, assessment tasks requiring verification of AI outputs, process evidence and clear institutional rules on ethical AI use.