New Opportunities for Visualization & Data Exploration in the Age of AI

Presentation Date: Monday, August 10, 2026

Location: 2026 Northeast Star and Planet Formation Meeting at Boston University

Abstract: AI is getting remarkably good at helping with scientific data analysis and visualization—but what does that actually mean in practice? I’ll show examples ranging from exploratory visualization to a new analysis of the velocity field of the nearby galaxy M83, where AI helped us do things that had resisted both standard algorithms and considerable human effort. ChatGPT and Claude helped fit complex structure, suggested useful analysis methods, generated working glue sessions and interactive visualizations, and helped compare emerging results with the literature. They also made mistakes—sometimes while confidently describing a result that was visibly different from what their own code had produced. The experience has convinced me that looking carefully at the actual visualization is more important than ever. AI can now generate analyses faster than even an expert can absorb them, making the human role less about doing every calculation and more about noticing what is wrong, what is surprising, and what might actually be telling us something new about the Universe.

Slides are available in Keynote and PDF formats on Dataverse, or as a quick compressed PDF download from this site.