Ashley Frith
Benedicte Knudson
Margot Lin
Katie McIntyre
Hang Wang
Project Team Members: Ashley Frith, Benedicte Knudson, Margot Lin, Katie McIntyre, and Hang Wang
Time Frame: September to December 2023
Tools: Qualtrics, Tableau, Dedoose, Miro, Figma, Adobe Creative Suite
PROBLEM SPACE
After our initial research, the need for an LLM application builder that supports user data upload and contributes to AI literacy, democratization, and accessibility efforts became increasingly clear. This would greatly benefit novice users as they build applications that employ large language models.
How might we support less-experienced users in crafting custom LLM applications using their own data?
RESEARCH QUESTIONS, METHODS, AND FINDINGS
Click to advance...
From our findings we generated...
DESIGN IMPLICATIONS
DESIGN PROCESS OVERVIEW:
1. Moodboards and inspiration
2. Sketches
3. Lo-fi wireframes
4. Hi-fi wireframes
5. User testing
6. Feedback
SKETCHING
The artifacts I produced at each iteration. Began with paper sketches, moved to digital sketches, and then lo-fi wireframes, receiving and incorporated feedback at each iteration.
APPLICATION FEATURES
Mid-fidelity
HIFI WIREFRAME
Try it out!
EVALUATION
We performed cognitive walkthroughs with experts and usability testing with novice users to see how well sandbox.ai matched with user mental models, understand how steep the learning curve was for novice users, test the compliance with design requirements, and test the prototype's accessibility. During the tests, we asked our users to think aloud as they walked through the application and completed tasks.
STAY TUNED!
