Adapting Collegiate Data Science Curricula for LLMs, Generative AI, and Beyond: Perspectives from Career Data Scientists

AI-based tools, generative AI, and large language models (LLMs) are transforming what it means to work and learn within the data science profession. We argue that higher education instructors should consider if, how, and in what ways to adapt data science curricula and pedagogy so that students who aspire to become data scientists are aware of how AI-based tools are integrated into and influencing the profession. To gain a sense of what modifications should be made to data science curricula (and why), we interviewed data science professionals across a variety of industry domains. We asked them to describe their career preparation, current work, and how employers expect them to use artificial intelligence tools and LLMs. Our findings suggest a general consensus that frequent human-AI interaction is necessary. A growing presence of AI tools in the workforce signals a stronger shift toward data pipelining, stakeholder communication and collaboration, and knowledge of evaluating AI output with epistemic expertise. Data professionals posed suggestions about how higher education might incorporate AI, generative AI, and LLMs into curricula, assessment, and practice, although some recommendations (e.g., learning to code with AI agents) would require a careful re-examination of the epistemic goals of data science by instructors. We illustrate how we took up professionals’ suggestions and adapted an undergraduate data science course designed for AI-integration in response. We conclude with a reflection on data science instructional challenges and potential roles that industry-academia collaborations might play in the AI era.