HCI research on AI agents
Improve agent-human communication through research and design.
Overview ◑ω◐
This project explores the design requirements for improving human-agent communication.
As AI gains more autonomy, new interaction patterns introduce unexplored risks. However, current AI design guidelines mostly focuses on non-agentic generative AI.
This project is funded by the Center for Long-Term Cybersecurity (CLTC) and is currently in progress. The expected outcome is a research paper and a website with design recommendations.
Currently conducting HCI research with the team. Next, I'll lead design, prototyping, and testing.
Focus
Agentic AI, Trust, HCI
Status
In progress
(expected to finish in May 2025)
(expected to finish in May 2025)
Project type
Master's Capstone project
Teammate
1 UX Researcher
1 Product Manager
1 Data Analyst
How might we design to improve agent-human communication—ensuring clear goal communication, actionable feedback, and transparent behavior?
Process
This project is still brewing! Stay tuned ((^u^))
01
Literature review
• Define trust in AI
• Identify gaps and new research opportunities
02
Heuristic evaluation
• Assess design patterns and usability of Operator
• Identify opportunities & challenges in existing solutions
03 (We're here!)
User interviews & usability testing
• Understand user perception of trust and how they develop trust
• Identify factors that lead to overtrust or undertrust
04
Wireframe & iterative prototyping
• Brainstorm design solutions to improve trust in AI agents and mitigate risks
• Designs will be validated through iterative prototyping
