Valuescribe
This ongoing project is to improve design education at Utah State University by increasing students' awareness of their own values as designers. Valuescribe is an artificial intelligence we are developing that will cowrite design fictions with students and synthesize their design values. I conducted stakeholder interviews, user interviews, secondary research, wireframing, community-led design and usability testing to design our current iteration.
*This project is a part of the research lab of Ha Nguyen, PhD. The original idea for Valuescribe is their own.
Role
Research team member and primary UX designer
What I did
Qualitative analysis, case studies, chatbot design, and research session conducting
Result
AI prototype that encourages more regulative thinking.
Final chatbot prototype
Background
Value sensitive design is a practice founded on the idea that technology should account for human beings' values. We take a broad definition of value: what people deem important in their own lives. This project seeks to inject design education with a self-awareness of values, i.e., students learn about design and who they are as designers through gaining awareness of what they consider important.
To inject design with values, we utilize design fictions. These are short stories that take place in a future where a certain problem has been solved. We write these as designers to generate specific details about our design solutions, and approach with use cases in mind. Because characters of the story will interact with the solution, we must think specifically about how people will use solutions.
This project is associated with research at Utah State University analyzing the efficacy of design fiction cowriting on value-sensitive design learning.
Process
To inform the design of a cowriting chatbot, I did the following:
-
Stakeholder interview
-
User interviews
-
Literature review
-
Preliminary user testing
-
Wireframing
-
Prototyping
Early Wireframes
Interviews
Stakeholder Interview
In this case, the "stakeholder" of the process was the leading professor for this project at Utah State University. I interviewed the stakeholders to discover the expected solution and its users.
Key Takeaways
-
Two student groups: application and theory.
-
Outcome of solution: students integrate more advanced design concepts into their design processes.
-
Notes: need to see what the student is thinking as they are doing the activity.
-
Agency: student can reject everything the AI says.
-
Open-ended interaction or AI led?
Mind map from Stakeholder interview
User Interview
An instructor and student were interviewed to understand the two primary user groups.
Key Takeaways
-
ValueScribe needs to allow for instructor control.
-
Don’t be idealistic, most people are here to graduate and nothing more.
-
Student sees artificial intelligence as limited but lots of potential.
-
Student interested in doing design fiction if it is about a problem they are captivated by.
-
Student would want to start slow with AI, and would not integrate AI responses immediately.
Mind map from Instructor interview
Literature Review
I review previous research I had done with design students about their perceptions of AI.
Student perception of AI, data from AI Design 2023
Key Takeaways
-
Artificial dialogue comes in as the top student perception of AI from our previous study. We need to address this in our testing and iterations; is our dialogue natural?
-
Generic is another concern, how will we design the experience to directly address this concern?
Preliminary User Testing
I brought in students for a design fiction co-writing preliminary with ChatGPT. Participants were instructed on what design fictions are, what cowriting is, and then they cowrote a design fiction with the chatbot.
Key testing points:
-
Design fiction instruction: do students get what a design fiction is?
-
Cowriting interaction insights
-
Enjoyments and frustrations
Key Takeaways
What went well
-
Instructions and contextual information seemed ample enough for participants.
-
Examples from pop culture (Star Trek, Minority Report) were valuable.
-
Problem examples were well-liked and applicable to students.
What went poorly
-
AI continuously generated cliche settings and solutions.
-
Participants were overwhelmed in the beginning.
-
Unclear guidelines of when the story is complete.
Participant Frustrations
-
Overwhelmed without examples of problems to work on.
-
AI distracted from what participants wanted to write about, or created unliked ideas.
-
Cliche setting and ideas.
-
AI losing details of story/forgetting prompting (GPT consistently devolved into rewriting what was said).
-
AI kept moving away from the purpose of the task–kept writing a fiction, but not a design fiction.
Participant Enjoyments
-
Generating new ideas and perspectives.
-
Expanded horizons and thinking.
-
Fun to write a story.
Initial Wireframes
Results
*Active project, Figma file may not behave correctly.
This iteration of Valuescribe allows users to view their library, generate story context, write a story with Valuescribe, view a presentation of their design values, and to reflect on the experience.
Informed Design Attributes
Branding
Essence: Creativity, books, innovation, playfulness, personality tests, and fun.
Fonts: Helvetica (designer’s font, creativity) and Libre Carlson Text (often used in books, books)
Colors: Dark with neon. Speaks to innovation and creativity.
Imagery: Painted and loose, encourage playfulness and fun, used in combination with neon colors to create an innovative but fun feel.
Wrap-up
What would I do differently next time?
I would involve more people in user testing, contributing more findings for the design.
Next steps
Human-AI study-in progress
We are currently doing a human study on the emotional reactivity of co-writing with AI vs human beings. This will inform the design of Valuescribe to promote more natural and human-like collaboration.
Diary study-in progress
Users will take home a developed version of Valuescribe and complete design fictions over the course of several weeks. This will more closely mimic how actual users will eventually use Valuescribe.
Division of Labor
This project is a segment of work in Dr. Nguyen's research lab. The original idea for software was their own, as were several design choices.
Jake
-
Design research
-
Designing
Dr. Nguyen
-
Original idea
-
Flow of design
Collaborative
-
Research study writing