AI-Assist to AI-Innovate: A Developmental Progression Taxonomy of User Proficiency in Generative AI

The rapid integration of Generative Artificial Intelligence (GenAI) across educational and professional contexts has underscored the need to conceptualize GenAI proficiency. While existing AI and GenAI literacy frameworks primarily emphasize conceptual understanding, they often overlook the dynamic and developmental nature of GenAI. This study proposes a taxonomy that delineates five progressive levels of GenAI proficiency. Anchored in the Dreyfus model of skill acquisition and informed by the Structure of Observed Learning Outcomes (SOLO) taxonomy and the Visible Learning Three-phase Model, this taxonomy offers an understanding of the progression of user interaction with GenAI, based on the nature of the task and cognitive engagement. Developed through an inductive approach, the taxonomy is derived from empirical insights gathered during a co-design process involving 26 participants from diverse academic and professional backgrounds, as well as a synthesis of 26 conceptual and theoretical papers. The resulting taxonomy delineates five distinct levels of GenAI proficiency, ranging from AI-Assist to AI-Innovate-each representing a progressive level of human-AI interaction, from basic information retrieval to advanced knowledge creation and real-world application. The taxonomy acknowledges that users may operate at different proficiency levels depending on the nature of the task and the cognitive engagement. It is intended to serve as a practical instrument for GenAI users to assess and support the development of GenAI proficiency.