By Xiaoyi Yang (Northeastern University)
Information
The activity presents a structured assignment design that integrates generative AI into an upper-level data visualization course. Rather than banning or passively allowing AI, the approach makes AI use explicit, reflective, and comparative. Students complete targeted activities including: (1) exploring visualization concepts from interdisciplinary perspectives using AI, (2) comparing human-written and AI-generated code, (3) using AI to interpret complex code before improving it, and (4) analyzing common AI failure cases. AI becomes both a tool and an object of critique. The method emphasizes judgment, task abstraction, and design reasoning over syntax. By embedding AI into structured prompts and reflection requirements, students learn not only how to use AI, but when and why to use it. The activity is integrated into weekly homework assignments as well as an extra-credit "AI in Visualization" workshop in an upper-level data visualization course. AI usage is clearly labeled in both the homework and the workshop materials, and selected AI prompts are provided to guide students. When prompts are not provided, students are required to document the prompts they use, along with the corresponding AI outputs. In addition, students must provide critical reflections on the AI-generated responses. This approach is implemented at a large research university in an upper-level undergraduate data visualization course within a data science curriculum. Students come from diverse academic backgrounds, including data science majors as well as non-majors such as computer science, business, and biology students, resulting in varied levels of coding experience and preparation. Class sizes are typically large ranging from 60 to 100 students per section, with three sections offered each semester requiring scalable assignment structures that accommodate different learning paces. Because many students already use AI tools independently, this structured integration responds to an existing reality rather than introducing a new dependency. Early observations suggest that AI use often helps students overcome localized barriers such as unfamiliar functions or syntax rather than abandoning entire problems, supporting persistence and task completion. In addition, an end-of-semester informal survey will examine changes in students' confidence using AI, their perceptions of appropriate boundaries, and their learning strategies.