Non-automatable cognitive skills in higher education in the age of generative AI

Abstract

The emergence of Generative Artificial Intelligence (GenAI) has led to an urgent reconsideration of which cognitive and metacognitive capabilities must remain central to higher education curricula. However, there is limited synthesis of the skills that cannot be meaningfully reduced to, or outsourced through, automated text generation. The current review addresses a gap in bringing together conceptual, pedagogical, and assessment-focused evidence on the human cognitive work that should be protected, developed, and evidenced in GenAI-enabled learning. This is essential for curriculum and assessment redesign in contemporary higher education contexts. The current study presents a narrative review of a total of 24 studies examining cognitive and metacognitive skills that resist automation in higher education contexts. Searches were conducted in Scopus, ERIC, Semantic Scholar, and Google Scholar for literature published between January 2020 and December 2025. Across the studies, higher-order critical thinking emerged as the most frequently cited non-automatable skill, alongside authorial decision-making, contextual disciplinary judgment, ethical and epistemic agency, and metacognitive skills such as self-regulation and reflective thinking. Skills resisting automation were characterised by accountability, contextual reasoning, and second-order self-regulation. The literature recommended learning outcomes that emphasise higher-order thinking, AI literacy and transversal competencies, including academic integrity, collaboration, and leadership. Evidence indicated that active, collaborative and metacognitively oriented pedagogies, and scaffolded human-AI co-creation, support these skills more effectively than unguided GenAI use. GenAI's impact on non-automatable skills is contingent on instructional design. Structured, reflective integration can support critical thinking and self-regulation, whereas uncritical or fully outsourced use may suppress independent analysis. Higher education should therefore redesign curricula and assessment to make accountable human thinking visible.