What Can GenAI (Really) Do for Data Visualization?

Short report from a new panel we organized at Northeastern University.

Last Thursday, Paolo Ciuccarelli and I organized a new panel at Northeastern’s Center for Design. This is the second time we have organized a panel covering topics at the intersection of Data Visualization and AI. Last time, the scope was broader; the panel title was “Where Does Data Visualization Fit in an AI-driven World?” This time, we focused more specifically on the role of generative AI.

For the panel, we invited three guests. We had Vidya Setlur, Victor Dibia, and Pau Garcia. Vidya is the director of research at Tableau. She has been working at the intersection of NLP and visualization for the longest time, pioneering some of the early approaches of natural language to data visualization and then working on Tableau’s AskData. This tool allows using Tableau with natural language. Victor is a principal research software engineer at Microsoft Research, and he developed one of the first LLM-based data visualization tools called LIDA. Pau is the Director and Founding Partner of Domestic Data Streamers, a studio focusing on various types of data art and design projects, including projects that mix AI and visualization.

The panel started with the three panelists’ statements and followed with their answers to a series of questions from me and then from the audience. I won’t do a full recap; you can watch the recording here.

I’ll highlight a few things that stood out from the conversation.

All panelists highlighted a paradoxical aspect of AI's recent development: AI is enabling humans to think more smartly and faster. Instead of replacing human thinking, generative AI is often used to augment it.

Another aspect is the idea of democratizing access to programming and data tools. Before the advent of generative AI programming, it was confined to a very small fragment of the population, but now it’s possible for a larger fraction of people to have a foot in the door of this world. Even a small increase can have a huge impact on creativity and opportunities.

One last topic that I found fascinating is the ability of AI to support creativity. Pau had a great slide about divergence and convergence, claiming that traditional tools typically support convergent thinking, but GenAI tools, for the first time, support more divergent thinking because it’s way easier to try things out. Also, Victor had a great point about the ability of future AI systems to be more creative than they are now.

Our panelists raised many other interesting points, and I was really impressed by the quality of our conversation. It gave me a lot of food for thought and made me even more excited about the future possibilities in this space. I hope you’ll watch at least part of it. It’s worth your time!