As artificial intelligence (AI) continues to reshape how we live and work, this article argues that the most consequential force behind contemporary AI is not technological acceleration alone, but a gradual redefinition of what counts as data. Because computational systems ultimately operate on data, advances in AI are inseparable from the expanding scope of data they can represent and process. Beneath the rapid progress of large language models, multimodal systems, embodied agents, and world models lies a deeper shift: data increasingly extends beyond relationally structured records to include language, imagery, interaction traces, environments, and systems capable of continual learning and internal representation.This expansion challenges long-standing assumptions and missions within data science. Data are no longer confined to purposefully sampled, rigidly organized, and retrospectively analyzable artifacts; they increasingly take the form of expressive, evolving, and population-scale representations that strain conventional analytical frameworks and validation practices.Rather than treating AI merely as a technological advancement, this article examines it as an expansion of the analytical landscape in which data science operates. Using Generative AI (GAI) as a focal lens, it explores emerging divergences between structured and expressive data, inferential and generative modeling, and applied decision support versus foundational analytical inquiry. These tensions become especially visible in education, where questions about what students should learn, what may be delegated to AI systems, and what must remain grounded in human judgment are increasingly difficult to ignore. The goal is not to prescribe a single future for the field, but to offer a reflective examination of how these developments may shape the evolving missions of data science research, practice, and education in the age of AI.
Zhiwei Zhu (Mon,) studied this question.