Designing for Learning with Generative AI is a Wicked Problem: An Illustrative Longitudinal Qualitative Case Series

Abstract

Students continue their education when they feel their learning is meaningful and relevant for their future careers. Computing educators now face the challenge of preparing students for careers increasingly shaped by generative AI (GenAI) with the goals of supporting their learning, motivation, ethics, and career development. Our longitudinal qualitative study of students in a GenAI-integrated creative media course shows how this is a "wicked" problem: progress on one goal can then impede progress on other goals. Students developed concerning patterns despite extensive instruction in critical and ethical GenAI use including prompt engineering, ethics and bias, and industry panels on GenAI's career impact. We present an analysis of two students' experiences to showcase this complexity. Increasing GenAI use skills can lower ethics; for example, Pat started from purposefully avoiding GenAI use, to dependency. He described himself as a "notorious cheater" who now uses GenAi to "get all the right answers" while acknowledging he's learning less. Increasing ethical awareness can lower the learning of GenAI use skills; for example, Jay's newfound environmental concerns led to self-imposed usage limits that impeded skill development, and new serious fears that GenAI would eliminate creative careers they had been passionate about. Increased GenAI proficiency, a potential career skill, did not improve their career confidence. These findings suggest that supporting student development in the GenAI era is a "wicked" problem requiring multi-dimensional evaluation and design, rather than optimizing learning, GenAI skills, ethics, or career motivation individually.