RUME 28
2026
Alexandria, Virginia
Artificial Intelligence and Cognitive Demand During a Domain and Range Mathematics Task
Page: 495
This exploratory study examines the nature of students’ cognition during high demand mathematics tasks when an AI tool is available. Using video and screen recordings from 20 undergraduates, we analyze students’ cognitive demand on a task involving a nonroutine roller coaster application of domain and range. Our findings show wide variability in how students approach the task, how and how often they elected to use ChatGPT (half chose to use it, half did not), and the levels of cognitive demand that students exhibit in their solutions and thinking. Students that solve the task with high demand tend to grapple with its nonroutine aspects and challenge AI output, while students that solve with low demand give decontextualized prompts, struggle to use AI output, and/or trust AI despite hallucinations. We discuss implications that undergraduate mathematics researchers and educators might consider to preserve opportunities for students to experience high demand problem solving.