What Can Generative AI Do for Data Visualization? Opportunities and Production Challenges from the Field
Generative AI models encode broad world knowledge that make them particularly useful for data visualization tasks - from suggesting data exploration directions,data preparation, generating visualizations, accessibility descriptions, semantic enrichment among others - all based on natural language descriptions. While this technology helps democratize visual analytics, the road to productizing these capabilities in the real world, has challenges..
In this talk, Victor will discuss the set of tasks for which Generative AI has begun to manifest in visual analytics applications with practical examples from the
LIDA OSS library, explorations with
Multi-Agent Systems and features across Microsoft data products. We'll explore often less-discussed topics that emerge during productization, such as –evaluation metrics, reliability, costs and latency. Finally, Victor will highlight pressing open research questions that could shape the future of AI and data visualization.
Victor Dibia (PhD) is a Principal Research Software Engineer on the
Human-AI eXperiences (HAX) team at Microsoft Research. His work on
LIDA, an
open-source tool for automated visualizations and infographics using generative AI, has informed the design of features in core Microsoft products like PowerBI, Excel, and
Project Sophia, as well as numerous other efforts at Microsoft. Victor's recent work has also spanned generative AI agents, where he has helped create
AutoGen Studio, a no-code tool for prototyping and debugging multi-agent applications. His research has been published at conferences such as ACL, EMNLP, AAAI, and CHI, and has received multiple best paper awards. Victor holds a PhD in Information Systems from City University of Hong Kong and an MSc in Computer Science from Carnegie Mellon University. He previously worked as a Principal Research Engineer at
Cloudera Fast Forward Labs and as a Research Staff Member at
IBM Research.
