Teaching Data Visualization in the Era of AI: From Du Bois to Large Language Models

• In Event: Teaching and Learning Symposium Roundtable Session
  In Refereed Roundtable (90 minute): Table 3: Navigating AI and Digital Literacy

Abstract

In recent years, sociology has seen a reintegration of data visualization in scholarly work (Healy and Moody 2014). Data visualizations have strong roots in classical works by foundational scholars such as Du Bois (1900) and offer an efficient way to communicate findings to a variety of audiences. However, when creating visualizations, it is important to consider what the visualization conveys, whether it is the best type for the message, and whether it is accessible to different audiences. Precision and aggregation decisions shape the stories visualizations tell; without attention, they can mislead (Healy 2019). This paper details innovations in teaching data visualization in an era where artificial intelligence is increasingly part of research workflows (Abramson et al. 2026).

Learning Objectives. Students learn best practices in data visualization, become mindful of unintended messages conveyed visually, and develop skills for working with emergent technologies such as large language models (LLMs). The overarching goal is to help learners create engaging visualizations that are comprehensible, accessible, and analytically responsible.

Approach. We discussed W.E.B. Du Bois's data visualizations from the 1900 Paris Exposition (Battle-Baptiste and Rusert 2018) and recent interactive dashboards, showing that data visualizations can be creative and effective tools. We examined misleading visualizations. Students completed a "dueling data visualizations" assignment requiring them to produce competing visualizations from the same data, demonstrating how design decisions shape a visual's story. Finally, we conducted in-class exercises creating data visualizations using LLMs, combining live coding with distributed expertise to show the utility of these tools and the decision-making process involved.

Usefulness. These pedagogical strategies address a growing need as data visualizations become standard practice across the social sciences and across data types. Audience members will leave with concrete activities for improving data literacy and integrating AI tools into their own courses, applicable at both introductory and advanced levels.

Authors

• Jakira Silas, Rice University

• Corey M. Abramson, Rice University