Improving the Utility and Usability of Visualization in AI-driven Scientific Discovery

Abstract

Visualization plays a critical role in XAI due to its efficiency in organizing and communicating information. However, current visualizations are designed mainly for AI researchers and offer little to no support for domain experts such as biologists or clinicians. Failing to consider the perspective of users and the characteristics of the applied research domains can significantly undermine the utility and usability of visualizations and explanations. We discuss the challenges and opportunities about the utility and usability of visualization in AI-driven scientific discovery.

Citation

Q Wang, S L’Yi, N Gehlenborg. “Improving the Utility and Usability of Visualization in AI-driven Scientific Discovery”, OSF Preprints (2022). doi:10.31219/osf.io/hpd6u