Carnegie Mellon University - Human-Computer Interaction Institute
Research Assistant and Developer | May 2024 - Current | Pittsburgh, PA
Overview
I worked with Drs. Jason Hong, Ken Holstein, and Motahhare Eslami conducting research on auditing generative AI models. This research focuses on developing tools and methodologies to evaluate and compare text-to-image AI models to better understand their capabilities, limitations, and potential biases.
Key Contributions
- MIRAGE Development: Created MIRAGE, a software platform that facilitates side-by-side comparison of different Text-to-Image AI models, enabling researchers and users to audit and evaluate generative AI systems more effectively.
- Research Presentations: Presented 5 research posters at various academic venues, disseminating findings on AI model auditing and evaluation methodologies.
- Publications: First-author demo paper at AAAI HCOMP '25 and first-author demo paper at ACM CI '25, contributing to the growing body of research on generative AI transparency and accountability.
Research Impact
This work advances the field of AI auditing by providing practical tools for comparing and evaluating generative models. Although I joined this project during the summer as part of an REU, my contract has been extended to continue conducting research with Carnegie Mellon University remotely.