AI in Viz (Optional)

Lesson 26: AI in Viz

Using AI in visualization, including in imeplementation.

Objective

Offer a hands-on exploration of using AI in constructing data visualizations through two LLM-focused labs.

Outline

An optional extra lesson looking at using generative AI. We do this by re-constructing an example data visualization from one of the early lectures of the course (predation / prey dynamics in Isle Royale).

Since the original course

Since the original teaching of Stat 198, some things have changed.

• Generative AI has grown in popularity but so has resistance to it.
• Some of the ethical considerations are outside the scope of this class but remain relevant.
• Encourage looking at Gebru at https://satml.org/2023/videos/.
• Libraries like Sketchingpy have added llms.txt or similar to help aid use.

However, the rules and lessons haven't really changed.

New instruction

Lesson 26 goes beyond the original material of Stat 198 as an update to prior instruction. Specifically, new skills lab 6 offers two new tutorials.

• Tutorial 13 looks at Matplotlib and re-generating the graphic we examined early in lecture.
• Tutorial 14 introduces advanced techniques and branches to 14a (browser) and 14b (agents) for Sketchingpy workflows.
• Requires an AI assistant (we demonstrate with Claude).

Role of AI

AI does not provide a silver bullet for understanding or visualizing data.

• Limitations of language: it can be hard to describe what exactly is desired.
• Don't forget the role of iteration.
  • Avoid offloading responsibility to AI.
  • Consider how AI can afford faster iteration.
• May provide new capabilities for user experience.

Take Aways

AI can help craft data visualizations but may serve more a role in accelerating implementation. However, it may be less helpful in design, especially as it can be difficult to describe in words what is desirable.

• Explore if AI is right for you, it's OK not to use it.
• LLMs can provide a quick visualization.
• Use tools like LLMs.txt to access libraries.
• Remain vigilant in iteration and use LLMs to try more ideas faster.

Citations

[1] Anthropic, "Claude," Anthropic, 2024. Available: https://www.anthropic.com/claude

[2] J. Willison, "llms.txt," llmstxt Project, 2024. Available: https://llmstxt.org/

[3] A. Pottinger and Sketchingpy Contributors, "Sketchingpy," Sketchingpy Project, 2025. Available: https://sketchingpy.org/

[4] J. D. Hunter, "Matplotlib: A 2D Graphics Environment," Computing in Science & Engineering, vol. 9, no. 3, pp. 90-95, 2007, doi: 10.1109/MCSE.2007.55.

[5] Isle Royale National Park Michigan, "Wolf & Moose Populations." National Parks Service, Mar. 29, 2024. Available: https://www.nps.gov/isro/learn/nature/wolf-moose-populations.htm

[6] T. Gebru, "Eugenics and the Promise of Utopia through Artificial General Intelligence," IEEE Conference on Secure and Trustworthy Machine Learning (SaTML), 2023. Available: https://satml.org/2023/videos/

License

This lesson is part of Interactive Data Science and Visualization and is released under a CC-BY-NC 4.0 license.