With the rapid adoption of Artificial Intelligence (AI) tools in personal, professional, and academic life, AI literacy is becoming increasingly important for critical engagement with media and technology. While much research has been done on AI literacy in K-12 education, there is a lack of insight into disciplinary differences in higher education, i.e., how awareness, usage, evaluation, and ethics of AI differ between science, technology, engineering, and mathematics (STEM) students and social science students. It is also unclear what role gender and prior AI experience play in this context, and what other factors there may be related to AI literacy. To address this lack of clarity, a link between computational thinking skills (CTS) and AI literacy has been proposed, as the CTS provide a versatile, lifelong learning skill set that is intended to facilitate the navigation of an AI-driven educational landscape. 

The study compared AI literacy and CTS between STEM and social science students, explored the relevance of gender and prior AI experience in relation to these two skill sets, and investigated whether higher levels of CTS are associated with higher levels of AI literacy. The analyses were carried out from a digital divide perspective, acknowledging structural inequalities in terms of generational vulnerability, discipline, and gender, and aiming to contribute to social sustainability, especially SDG 4.7, by recommending equitable educational interventions. A survey was conducted with N = 380 undergraduate students from a university of applied sciences, combining existing, validated questionnaires with an open-ended experience reflection, resulting in quantitative and qualitative empirical material for the analysis.

Key findings revealed no statistically significant difference in AI literacy across disciplines, although the qualitative analysis uncovered nuanced differences. The difference in CTS and the association between CTS and AI literacy were significant, but with small effect sizes (< 0.10), lacking practical implications. ‘Ethics’ and ‘algorithmic thinking’ were the weakest dimensions, requiring educational attention. Moreover, an asymmetric pattern of frequent AI use and little formal AI training was found, which is problematic for students, higher education institutions, and society at large. Increased efficiency was overshadowed by concerns about overreliance on AI tools, which poses a risk to equity in the absence of efforts to strengthen autonomy. Also, gender emerged as a relevant factor, reflecting discipline-specific stereotypes that need to be addressed. Overall, the study highlights the need for formal, interdisciplinary, and gender-sensitive AI literacy training to strengthen equitable AI adoption in higher education.