An Empirical Study Into Higher Education Students’ Critical Thinking Towards AI-generated Output

Previous research has revealed a decrease in newly enrolled higher education student’s critical thinking ability. In addition, studies have reported on the negative implications of artificial intelligence (AI) for students’ cognitive abilities. When comparing studies that have investigated the relationships between critical thinking, AI-usage and the intelligence domains ‘fluid and crystallised intelligence’, contradicting findings arise. These findings provoke further examination of their interplay. This study aims to bridge this gap by investigating the relationship between German higher education students’ (n = 353) intelligence domains (fluid and crystallised) and their ability to critically assess AI-generated output for their studies on its accuracy, reliability, and applicability (called AI-CT). The collected data is part of a large-scale online survey called ‘Fachkraft 2030’, in which the economic, educational, and general living situations of students and young graduates residing in Germany are asked and additional cognitive tests are conducted. Hierarchical weighted least squares regression analyses have been conducted to investigate both fluid and crystallised intelligence in relation to AI-CT. Influential factors on AI-usage – age, highest academic accomplishment, intended study purpose of GenAI LLMs, and the Big Five personality traits – have been included as control variables. The results showed that after controlling for these factors, crystallised intelligence remained significantly positively associated with AI-CT (B = .21; p < .001), whereas fluid intelligence was no longer significantly positively associated with AI-CT (B = .09; p = .069). The study findings contribute to the understanding of critical thinking among students and bring about implications for educational practices.