Renee Kizilc presented research on student use and misuse of generative AI in academic settings. The first study, conducted with colleagues Igor Shirikov and Smernov, surveyed nearly 100,000 undergraduate students across 20 research-intensive universities and found that 37% use generative AI regularly for schoolwork, with STEM students using it more than non-STEM students. The study also revealed that 9% of students misuse AI in ways they know are not allowed, with misuse increasing to 26% among daily users. The research showed significant gender and racial gaps in AI usage, with male-identifying students and white/Asian students using AI more frequently than female-identifying students and underrepresented racial/ethnic students. The second study analyzed admissions essays from an engineering college and found that LLM usage in essay writing increased significantly after ChatGPT's release in 2024, with lower SES students using AI more extensively but facing higher penalties in the admissions process. Renee concluded by recommending discipline-specific approaches to addressing AI in assessment, including controlled settings, clear task-level norms, and redesigned tasks that incorporate AI use while promoting critical thinking skills.
Register to view a short clip of the research presentation here.
