Interviews
Once we reworked our project idea, we started the emphasis stage. At this point, we wanted to better understand our potential users by knowing their goals, needs, and challenges. We can better design and plan our application if we know more about our users. We started by developing interview questions. We identified two potential groups, those that exercise consistently and those that don’t. As a class, we actually discussed our questions and found some flaws. A few were leading questions and some others sounded accusatory. As a group, we rephrased several questions. Then we each conducted three interviews for a total of nine. Each of my interviews went fairly well. My first interview was very brief; it contained helpful information, but I didn’t receive a lot of descriptions for their answers. The other interviews were much better, they took on a more conversational tone.
Analysis
After gathering our own interview data, we all used ChatGPT to analyze and break down our information. It was my first time to really utilize AI to such an extent. We learned how to engineer prompts so that when we gave it to the interview transcripts, it could create empathy maps. For example, we told ChatGPT we had interview transcripts, to act as a UX researcher, generate empathy maps, and give the response with relevant quotes from the transcripts. It took some time, but it gave a response that utilized actual interview quotes, broken down into says, thinks, feels, and does. I made sure to check if the quotes were really from the interview. We then utilized ChatGPT again, but instead we told it to conduct a thematic analysis. Using a similar prompt as before, we gave transcripts and asked for it to generate a word cloud. It took time to analyze, but it came up with several themes prevalent to the interviews. Since we all did our own analysis, we came up with different themes and insights. As a group, we compared each list of themes to find commonalities and develop a list better suited for our project. We developed themes relating to motivation, notifications/reminders, tracking/accountability, and time. We learned that users could benefit from developing intrinsic motivation, adjustable intensity, notifications that promote growth instead of straightforward goals, and a system that can account for a busy schedule. In general, everything could use a level of personalization.
Reflection
Overall, AI as a tool really streamlined our process. Instead of spending an hour or more reviewing interview data, it was done in minutes. However, sometimes it would prioritize certain words, and we would have to prompt it to exclude them. It sometimes gave a list of information that was too much. Admittedly a lot of the information was good. It would provide context and breakdowns for everything. However, if we only needed themes, it would also break down user groups and more. With that said, it was really helpful. And if there was too much information, we could prompt it for less