Balancing Support and Deskilling

Abstract

Context: Generative artificial intelligence (AI) offers accessible task support, but the implications for independent learning remain uncertain. This paper examines perceived support, possible cognitive offloading, and self-regulated learning in vocational education and related higher education settings. Approach: A structured narrative synthesis examined six unpublished empirical bachelor’s and master’s theses completed at the University of Hamburg between 2023 and 2025. Four studies concern vocational education and two provide complementary higher education perspectives. Surveys, a task comparison, an uncontrolled pre-post study, and qualitative accounts were compared while retaining their methodological differences. Findings: Surveys reported substantial AI familiarity and use among respondents, perceived benefits, and concerns about uncritical reliance. In a small task comparison, AI-assisted responses more frequently met the completeness criterion, whereas responses produced without AI more frequently met the study’s transfer and independent-wording criteria. A vocational intervention found increases in self-reported self-regulation, motivation, and self-efficacy; its design cannot isolate an AI effect. Qualitative findings highlighted difficulties in organising problem-solving and evaluating understanding. Conclusions: Perceived usefulness, task-product quality, and independent competence are distinct outcomes. The studies do not establish long-term deskilling or a causal role of self-regulation in AI effects. The synthesis motivates instructional approaches that combine AI assistance with explanation, verification, and independent application, and identifies priorities for controlled and longitudinal research.