Study of AI use by undergrads revealing disparities in access – and in cheating (news.berkeley.edu)

A comprehensive study led by Igor Chirikov from UC Berkeley highlights significant disparities in AI usage among undergraduate students, revealing how socioeconomic factors impact both access to and the misuse of generative AI tools. The study, involving over 95,000 students across 20 research universities, found that approximately two-thirds of students engage with AI, but nearly 9% admitted to using it for cheating, particularly in non-STEM disciplines. The findings underscore a worrying trend where students from low-income backgrounds, racial minorities, and women utilize AI less, potentially putting them at a disadvantage in a workforce increasingly reliant on AI proficiency. The implications of these findings are profound for the AI/ML community and education as a whole. Chirikov emphasizes that blanket bans on AI tools are not effective and could hinder students' learning. Instead, the study advocates for universities to rethink assessment methods and incorporate AI literacy into curricula. With generative AI evolving rapidly, educational institutions must adapt to ensure all students can develop necessary skills while navigating the challenges posed by AI. This research not only signals a need for more tailored approaches to academic integrity but also raises important questions about equity in educational resources and opportunities in the age of AI.