Bias Detective: Improving understanding of bias through an integrated mini-game
Role
Product Manager,
Designer, Researcher
Team
Kai Shaw, Anna Shi,
Ye Wei, Aundre Watley
Timeline
Feb-Apr 2025 (3 months)
What is Bias Detective?
For my course "User-Centered Research and Evaluation", we were prompted to empower everyday users to uncover algorithmic bias in gen-AI platforms.
If users don't have a fundamental understanding of bias itself, how can they be expected to uncover algorithmic bias? This is why we created Bias Detective: a short engaging mini-game where players identify and learn about different forms of bias.
MY ROLE
I managed my team by assigning tasks, reporting progress to instructors, and ensuring our project aligns with user needs. I also facilitated all necessary user research and design tasks.
How did we develop our Research Question?
To brainstorm, we utilized the reverse assumptions method to break out of our conventional thinking patterns.
Reverse Assumptions
Using our ideas, we brainstormed concrete research questions and voted on our top three research questions.
Potential Research Spaces
We ended up going with the following question:
How might we elevate or improve college students' understanding of algorithmic bias within popular gen-AI platforms like ChatGPT?
Why a Mini-Game?
Once we had a solid understanding of our research questions and user needs through user interviews, we rapidly generated concrete solutions to our question.
Crazy 8's
After voting on our favorite ideas, we storyboarded out each concept and tested each one with participants.
One of our storyboards
We ended up going with the mini-game due to user feedback and user needs — specifically because we recognized that users want quick, engaging educational resources, so solutions such as lectures or videos wouldn't be ideal.
The Design Solution
Bias Detective is a short, engaging mini‑game that kicks off your ChatGPT session. Through four scenarios, users get instant feedback explaining how bias manifests as stereotypes or unwarranted generalizations in the output. By practicing with the given scenarios, users become more conscious of hidden assumptions and evaluate GenAI output with a more critical lens throughout the session.
Wrapping Up: Takeaways & Reflection
At the start of UCRE, we were asked to create three concrete learning goals (shown to the right). After finishing up my project, I want to reflect on what I learned in UCRE.
What i learned
There's an endless sea of HCI research methods to explore and pull from.
One of my goals was to expand my knowledge of HCI research methods to inform my designs. In UCRE, I learned many methods I'd never heard of before: Speed Dating, Parallel Prototyping, Crazy 8's to name a few.
Algorithmic bias in Gen-AI systems is a complex, diverse topic.
Since our project theme was tied to gen-AI bias, lectures and discussions provided me with a greater understanding of gen-AI bias and how to approach decreasing it. Thus, I have a greater understanding of how to apply AI ethically.
Team conflicts are impossible to work through without open communication.
Throughout the project, I felt tension with my teammates because I would always have to finish all the work on my own; there was lack of communication in general. However, I never expressed my concerns to my teammates, so resentment built up. While I didn't fully achieve my goal of working effectively in teams in UCRE, I have concrete next steps I can apply to future group work.
