About the Project
This project transforms AI onboarding from a static document or slide deck into an interactive, hands-on learning program focused on stroke rehabilitation video assessment. The project aims to help users develop the skills needed for Human-AI collaboration: recognizing where AI is reliable, identifying common failures, interpreting supporting evidence, and making adjustments that improve outcomes.
The intervention follows a U-C-I (Understand-Control-Improve) framework that introduces practice-on-failure pedagogy, actionable explainability with safe levers for edits, risk-aware previews of changes, and keeps human rules and AI behavior synchronized. The onboarding experience is designed to be measurable, so we can assess what users learn and whether it persists over time.
The project will deliver a validated onboarding toolkit that improves calibrated reliance, along with reusable literacy measures and UI/human-AI interaction patterns applicable beyond rehabilitation. The project will contribute a rigorous foundation for human–AI collaboration methods that support safer and more effective AI adoption while improving AI-assisted decision-making.
Research Impact: This project develops measurable, governed human-AI collaboration methods that enable safe and effective AI adoption in real-world decision workflows.
Theme: Open Category
Principal Investigator(s)
ResWORK Fellow, LEE, Min @ SCIS