Assistant Director, University Bursar, University of Illinois System Student Money Management Center
Generative AI is quickly being adapted to research activities, but we are still in the early phases of learning its limits. Exploring the “light side” and the “dark side” of this force as a qualitative research tool can help scale data collection efforts. A non‑programmer adapted an open‑source, Python‑based AI chatbot and spent several months comparing OpenAI and Anthropic APIs to identify the most consistent option for interview tasks. When implemented in research, the chatbot demonstrated several strengths. In one instance, it autonomously recovered a lost connection during an interview, preserving data while limiting demands on the participant. In another, it continued an interview with a participant who responded exclusively in Korean even though it produced only English text. The system also revealed weaknesses, such as repeating questions despite coded safeguards intended to prevent participant fatigue. Join to uncover the opportunities and pitfalls of using AI to scale qualitative data collection.