New study reveals cross-cultural divide in student views on generative AI ethics

A new cross-cultural study published in ACM Transactions on Computing Education reveals stark differences in how Canadian and South Korean computing students perceive the ethicality of generative AI (GenAI) use in assignments. Researchers Brian Harrington, Irina Zlotnikova, Gayathri Nadarajan, and Samuel Ekundayo surveyed students at a Canadian and a South Korean university in Fall 2024, presenting scenarios varying the amount of AI-generated code (e.g., 10% vs. 80% of a project), the nature of AI assistance (e.g., debugging vs. writing from scratch), and whether the professor explicitly allowed AI. Using Mann-Whitney U tests, they found statistically significant differences: Canadian students were consistently more likely to view GenAI use as both unethical and a violation of institutional policies, even though both universities had functionally identical honor codes.

The study's analysis using Hofstede's cultural dimensions framework suggests that cultural factors—particularly Canada's higher individualism and lower power distance versus South Korea's collectivism and higher uncertainty avoidance—profoundly shape ethical reasoning. The amount of AI-generated code incorporated into assignments emerged as the strongest predictor of ethical judgments across both groups. The authors argue for culturally responsive AI policies in global education, noting that a one-size-fits-all approach to academic integrity may unfairly penalize students from cultures with different norms around collaboration and authority. The findings highlight the need for nuanced guidelines that respect local cultural contexts while maintaining fundamental academic honesty principles, urging universities to engage in ongoing cross-cultural dialogue as GenAI use becomes ubiquitous in computing education.