Interactive Discovery and Exploration of Visual Bias in Generative Text‐to‐Image Models

Information

• Publication Type: Journal Paper with Conference Talk
• Workgroup(s)/Project(s):
  • Joint Human-Machine Data Exploration
• Date: June 2025
• Journal: Computer Graphics Forum
• Open Access: yes
• Location: Luxembourg
• Lecturer: Johannes Eschner
• Article Number: e70135
• ISSN: 1467-8659
• Event: EuroVis 2025
• DOI: 10.1111/cgf.70135
• Pages: 20
• Publisher: WILEY
• Keywords: Visualization, Bias, Artificial Intelligence

Abstract

Bias in generative Text-to-Image (T2I) models is a known issue, yet systematically analyzing such models' outputs to uncover it remains challenging. We introduce the Visual Bias Explorer (ViBEx) to interactively explore the output space of T2I models to support the discovery of visual bias. ViBEx introduces a novel flexible prompting tree interface in combination with zero-shot bias probing using CLIP for quick and approximate bias exploration. It additionally supports in-depth confirmatory bias analysis through visual inspection of forward, intersectional, and inverse bias queries. ViBEx is model-agnostic and publicly available. In four case study interviews, experts in AI and ethics were able to discover visual biases that have so far not been described in literature.

Additional Files and Images

Weblinks

• live demo
  Live demo of the ViBEx application
• Entry in reposiTUm (TU Wien Publication Database)
• DOI: 10.1111/cgf.70135

BibTeX

@article{eschner-2025-ide,
  title =      "Interactive Discovery and Exploration of Visual Bias in
               Generative Text‐to‐Image Models",
  author =     "Johannes Eschner and Roberto Labadie‐Tamayo and Matthias
               Zeppelzauer and Manuela Waldner",
  year =       "2025",
  abstract =   "Bias in generative Text-to-Image (T2I) models is a known
               issue, yet systematically analyzing such models' outputs to
               uncover it remains challenging. We introduce the Visual Bias
               Explorer (ViBEx) to interactively explore the output space
               of T2I models to support the discovery of visual bias. ViBEx
               introduces a novel flexible prompting tree interface in
               combination with zero-shot bias probing using CLIP for quick
               and approximate bias exploration. It additionally supports
               in-depth confirmatory bias analysis through visual
               inspection of forward, intersectional, and inverse bias
               queries. ViBEx is model-agnostic and publicly available. In
               four case study interviews, experts in AI and ethics were
               able to discover visual biases that have so far not been
               described in literature.",
  month =      jun,
  journal =    "Computer Graphics Forum",
  articleno =  "e70135",
  issn =       "1467-8659",
  doi =        "10.1111/cgf.70135",
  pages =      "20",
  publisher =  "WILEY",
  keywords =   "Visualization, Bias, Artificial Intelligence",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2025/eschner-2025-ide/",
}