Rethinking AI Literacy Education in Higher Education: Bridging Risk Perception and Responsible Adoption

Students struggle to recognize risks in realistic scenarios: While 139 students rated abstract AI risks as concerning when explicitly stated, they failed to identify those same risks when embedded in practical, scenario-based situations—suggesting a critical gap between theoretical knowledge and real-world risk recognition that could undermine responsible decision-making in actual AI development work.

Higher risk awareness doesn't guarantee cautious adoption: The study found a paradoxical "risk underappreciation" among AI-specialized students, who simultaneously demonstrated elevated awareness of explicit risks yet showed greater willingness to adopt AI systems—indicating that education focused solely on risk awareness may be insufficient for promoting responsible adoption practices.

Gender differences persist despite technical training: Although technical education narrowed gender gaps in risk awareness overall, male students consistently reported higher willingness to adopt AI despite similar risk perceptions, suggesting that factors beyond technical knowledge influence adoption decisions and may require targeted interventions.

Abstract warnings are less effective than concrete examples: Students responded much more strongly to explicitly stated, concrete risks than to abstract or scenario-embedded ones, implying that AI literacy programs should prioritize case studies and real-world examples over general risk discussions to meaningfully shape practitioner behavior.

Current AI education may not adequately prepare responsible practitioners: The disconnect between risk awareness and adoption willingness reveals that existing higher education approaches may be insufficient for cultivating the "ethically informed and socially responsible AI practitioners" needed for safe AI development and deployment.