Abstract
Generative AI exploded onto the scene of higher education in the past few years, but especially in Fall 2025, when the first students who used the tools, such at ChatGPT or Gemini, in high school enrolled in introductory college courses, including introductory computer science courses, that both built toward teaching how to write your own generative AI, and also faced a legitimacy crisis were many existing assignments and problem sets were trivialized by generative AI tools. But by understanding both AI and undergraduate education, we can transpose our teaching techniques to (1) allows students to use these tools without (2) trivializing course content. We present Wuphala, a graphics generation assignment intended as a final project for a computing course. It relies on prompting students to generate certain graphics and visuals from plaintext descriptions that can only be unambigiously resolved with the assistance of reference images. We found AI agents incapable of completing this assignment, which mapped exactly onto the existing learning objectives of our introductory computing finals, and only changed how and where graphics were incorporated into the assessment. Additionally, through the selection of images and their arrangement, we highlight the global impacts of the usage of AI frameworks, with the Wuphala project being inspired by the plurinational socialist Bolivarian movement and its relation to cobalt mining in South America. By specifically linking imagery to humanity, we create a cohesive, AI-impossible, introductory assignment.
Details
Presentation Type
Paper Presentation in a Themed Session
Theme
KEYWORDS
Artifical Intelligence, Plurinationalism, Internationalism, Pedagogy, Machine Learning, Generative AI, Computing
