Designing AI-Resistant Assessments in Higher Education: A Systematic Literature Review

HRMARS - The rapid proliferation of generative artificial intelligence (AI), particularly ChatGPT, has disrupted conventional assessment practices in higher education. AI has enabled common tasks like reports and essays to be completed with minimal human authorship, raising pressing concerns about academic integrity, validity, and pedagogical relevance. This systematic literature review synthesizes recent empirical and conceptual papers to examine three interrelated questions: how AI challenges conventional assessment, which design strategies show promise in mitigating misuse, and where significant gaps remain in current research. Drawing on a PRISMA guided review of 28 peer reviewed studies published between 2022 and 2025, the analysis identifies five dominant themes; the structural limitations of traditional assessments, the shortcomings of AI detection technologies, AI resistant assessment design principles, the importance of authentic and process oriented assessment and the need for establishing transparent institutional policies. Findings indicate that relying on ChatGPT detection technologies is insufficient, while authentic, process-based, and AI-integrated approaches are more practical and sustainable. The review argues that the rise of generative AI should not be viewed merely as a threat, but a catalyst for revising assessment designs in higher education.