Before this class, I never fully understood nor consciously thought about how much planning and consideration have to go into designing a product. I also always knew how much importance psychology holds in every aspect of life, but never consciously thought about implementing specifically to this extent for the products—from planning to designing.
I did this work with these experiences: In this class, I really loved working with my team members. I always enjoy working in teams, but I think my fellow members were amazing, smart, funny and pulled their own weights. I also really enjoyed finally learning to see what vibe-coding was like through that one session in class as well. However, the numerous requirements for each task due each class section was overwhelming for me—especially when it required going back to search into old assignments details to recall certain terms/ examples. Recently with graduation approaching as SymSys AI track student, I’ve been trying to think more often of work beyond school, and I really wanted to do more project classes that stepped a bit away from NLP, especially if I want to dabble into creating and releasing a product of my own along the line. Therefore, this class really helped me with that. Reflecting back on this class, even if I were to forget and when I do forget the specific details of everything I’ve done in this class, my biggest takeaway from this class even 10 years from now would be, “Just because I think I’ve considered every possible possibilities will never mean I’ve not missing anything. Test. Experiment. Research. Backtrack. Improve.” One thing I definitely do wish that went differently in this class is that I wish we started our prototype earlier—I feel like the class was pretty rushed towards the end, especially in weeks 9 and 10. Therefore, I’d recommend starting it maybe a week or two earlier, so students can further flush out their products and work on feedback, even with the assistance of AI.
Ethical Considerations: One big ethical consideration that stood out to me the most throughout our project was the line between nudging and manipulation. Since we wanted to encourage our users to be more intentional with how they prompted LLMs, especially in academic contexts, we used nudges like scoring and gamification methods. I think these mechanisms count as acceptable nudges because they were implemented in a manner where our users’ autonomy to prompt whatever they liked still remind while they were also able to choose different difficulty levels based on how they feel about the score additions or reductions—furthermore, they were able to turn our product at any point in their sessions with the LLMs. At the same time, this was the ethical consideration that stood out to me the most because the users, no matter how willing they were when they first started using our product may still start to feel judged or pressured. Our feedback system after every input as well as the leaderboard format may instill those feelings. At this point, our product could start to feel manipulative to them since they may start to think that they are being restricted in how they can prompt the LLMs.
Now I think that designing a product is not about its functionality or appealing visuals. It is also about continuing to question our product’s effects on people even long after our product’s release. Next time I am faced with a similar situation, I will stress-test a lot more and prototype earlier, keeping in mind each ethical tradeoff from past experiences more since the beginning while staying open to change based on feedback and stress-testing.
