The impact of artificial intelligence on task performance and decision-making: Empirical evidence on generation Z

Abstract

This study examines how generative artificial intelligence (AI) reshapes task performance, decision-making, and evaluative judgement in higher education assessments, with a focus on emerging human-AI assemblages among Generation Z university students. A controlled three-stage scenario-based experiment was conducted with the same cohort of students of business and economics, comparing a baseline session (no AI), independent reasoning (no AI), and identical AI-assisted conditions. Participants completed tasks involving situational judgment, quantitative reasoning, and short written responses. Results reveal that AI access increased average performance but markedly compressed score variance and reduced internal reliability, undermining the assessments' diagnostic capacity to differentiate independent abilities. Qualitative findings indicate that students perceived non-AI conditions as more cognitively effortful and educationally valuable, with AI shifting agency toward tool management and oversight. Together, these results highlight how AI redistributes agency in assessment, raising questions about responsibility and validity in sociotechnical contexts. Based on these insights, the study recommends hybrid assessment designs that separately evaluate independent reasoning and AI-augmented performance, incorporating reflective components to render distributed agency visible and preserve evaluative judgement.