Data-related practices for creating Artificial Intelligence systems in K-12

Computer science curricula have started to include competencies related to artificial intelligence (AI) in K–12 education. However, before introducing a new topic into the classroom and suggesting competencies, it is essential to identify the central practices of the discipline. In the following research, we focus on identifying practices related to data, as current school curricula significantly underestimate the role of data, and understanding how data is processed is a key to understanding how AI systems function. We examine the theoretical literature on practices applied to data when creating AI systems, map the practices in a process model, validate the results of the mapping with domain experts, and contrast the results with current AI curricula for school students. The contribution of this work is a process model that summarizes data-related practices for AI systems built with machine learning, is comprehensively domain-embedded, and is aligned with K–12 education. Computer science educators can use it as a blueprint for defining competencies and designing learning arrangements that aim to enable students to create and understand AI systems.