MIRAGE: Multi-model Interface for Reviewing and Auditing Generative Text-to-Image AI
Conference on Human Computation and Crowdsourcing (Works-in-Progress) (HCOMP), 2024
MIRAGE is a web-based auditing workflow that allows participants to contrast outputs from multiple text-to-image (T2I) models on a single canvas. By centering lived experience and supporting side-by-side comparison, MIRAGE helped participants uncover subtle biases in generative models during a preliminary study. The interface offers a structured path for non-AI experts to document and submit model feedback that can inform safer, more inclusive generative AI systems.
@inproceedings{maldaner2024mirage,
title={MIRAGE: Multi-model Interface for Reviewing and Auditing Generative Text-to-Image AI},
author={Maldaner, Matheus Kunzler and Deng, Wesley Hanwen and Hong, Jason I. and Holstein, Ken and Eslami, Motahhare},
booktitle={Proceedings of the ACM Conference on Human Computation and Crowdsourcing (Works-in-Progress)},
year={2024}
}