Abstract

Artificial intelligence (AI)-based assessment systems are emerging as innovative tools to evaluate and enhance critical thinking and creativity in higher education. By leveraging deep learning algorithms, generative language models, and automated scoring techniques, these systems offer scalable, adaptive, and personalized feedback mechanisms aligned with 21st-century cognitive skill development. Despite increasing implementation, empirical evidence regarding their effectiveness remains fragmented. This systematic review synthesized the findings of original peer-reviewed studies assessing the impact of AI-driven evaluation tools on students’ higher-order thinking skills. Following PRISMA 2020 guidelines, comprehensive searches were conducted in PubMed, Scopus, and Web of Science. Inclusion criteria focused on university-level interventions evaluating critical thinking and/or creativity using AI-based assessment tools. Of 234 records identified, only three studies met all eligibility criteria for final inclusion. Data were extracted using standardized forms, and risk of bias was assessed with CASP checklists. The included studies applied diverse AI systems: a BERT-based short answer grading tool, a deep-learning-powered creativity assessment platform, and a GPT-3.5-based mock interview rubric. All reported strong correlations between AI-generated scores and expert human evaluations. Outcomes indicated that AI-based assessments reliably measured cognitive indicators such as inference, originality, communication clarity, and divergent thinking. However, ethical considerations, data transparency, and researcher-participant dynamics were insufficiently addressed across studies. AI-based assessment systems consistently demonstrated effectiveness in enhancing critical thinking and creativity among university students. This systematic review identified strong correlations between AI-generated evaluations and traditional human assessments, validating their reliability across cognitive domains such as inference, originality, and clarity. Despite ethical and methodological gaps in existing studies, the evidence supported AI’s potential as a valuable complement to human judgment in higher education. These findings directly address the research question and confirm that AI-based assessment tools, when implemented responsibly, can contribute meaningfully to the development of higher-order cognitive skills. ©: ©2025 The authors.