The objective of the Define stage was to take everything we learned from our interviews and secondary research and develop a clear direction. Rather than designing around a vague idea of “stressed college students,” we needed to define a specific user, understand their behaviors, and critically evaluate whether AI was actually necessary in our solution.

To support this process, I used ChatGPT as a tool, not to do the work, but to help us as a teammate and improve the quality of our decisions. One of the most useful moments came while we were building the user journey map. Our first draft mapped a student’s stress and scheduling process over a single week. It captured daily issues, but it didn’t fully reflect what we were hearing about burnout as a pattern that builds over time.

Prompt:
“How could we adjust the scope of our user journey map to better capture how burnout develops over time for college students balancing school, work, and personal commitments?”

AI Output:
“Consider expanding the journey beyond a short timeframe and instead mapping a longer academic period. Burnout often develops gradually through repeated stress cycles such as midterms, project deadlines, and finals. A broader scope could highlight cumulative fatigue, declining motivation, and reduced rest patterns.”

That suggestion pushed us to widen the scope of our journey map to better reflect the reality of student life. By mapping across a semester, we were able to capture patterns like declining rest, increasing stress and overwhelm, and how students often sacrifice sleep, meals, and social time as responsibilities become more important. This also helped us identify clearer design opportunities, specifically around schedule rebalancing when plans change and preventing burnout before it gets to be too much.

The biggest benefit of using AI in this stage was perspective. It helped us step back and see what our first draft didn’t capture, especially in how time pressure compounds. It also gave us a way to describe burnout as a long term experience rather than a one week issue. The main challenge was that AI outputs can be generalized and sound confident even when they aren’t specific to our data. We had to treat it like a brainstorming partner and filter everything through our interview patterns before changing anything in our work.

My team and I did the research, built the persona from real patterns, mapped the journey in detail, and completed the worksheet connecting user needs to data requirements. AI helped us question scope and strengthen our framing, but it did not define the problem for us. In the Define stage, AI was used as a tool for expanding our thinking, while the responsibility for accuracy, evidence, and design decisions stayed with us.