G.EM (Gen-AI for Empathy Building)
#UX Research #Figma #Emotion AI #Usability Test
G.EM is an emotion-aware intelligent system, envisioning an AI virtual mentor for user interviews. With the real-time emotion analysis and AI-powered insight generation, G.EM aims to help UX researchers increase the quality of qualitative research.
Timeline
Sep - Dec 2023 (3 months)
Project Type
MDes Thesis Project
Tools
Google Docs/Sheets, Figma, Hume.ai (Emotion AI), Otter.ai (Meeting Gen-AI), Visual Studio Code, Adobe Suite
My Contribution
User/Technology Research: I identified needs for UX researchers and explored the potentials of utilizing AI technology to solve the problem
UX/UI Design: based on research findings, I designed and prototyped G.EM as a platform to empower qualitative user interviews
Problem
Challenges in User Interview
User interviews, a common practice for every UX researcher, can be very challenging. In a short amount of time, researchers capture qualitative data through conversations, trying to understand their potential users' behaviors, preferences or lifestyle. The researchers need to master multiple skills to excel a user interview, which requires the most workload in the least time.
Research
Potentials of Emotion AI for User Research
I conducted research into Emotion AI technology to determine the potentials of utilizing it for user research. Emotion AI technology can detect, categorize and analyze expressional emotions from language used, speech prosody, facial expression, etc. The emotion data of user interviews can be very valuable for researchers to understand how to improve the research process.
Design
Gen-AI for suggestion generation
Taking emotions into consideration, I explored a design which uses Gen-AI to determine the quality of user interview conversation, and generate suggestions to improve the quality.
Prototype
Emotion-aware features
With the insights I gathered from research, I built a Figma prototype of G.EM’s features, from capturing to analysis to customization.
#1: Capture
Automatically take meeting notes during the interview, make it easy to rewatch the conversation details
Real-time capture the emotional data from voice burst, language use and prosody for analysis
#2: Suggest
Push easy-to-read tooltip on your chosen device to provide instant assistance during your meeting
Just enough to help you out when you're stuck without getting in the way of your conversations
#3: Insights
In-depth analysis after an interview session for reflecting and learning
Gives emotion-aware suggestions how you can improve this interview which is powered by Gen-AI
#4: Collect
Evaluate the suggestions with your own judgement, give feedback to AI model
Collect useful suggestions to Performance page for further practices, or fine-tuning AI model