Towards a Framework for Learning Content Analysis in K-12 AI/ML Education

Abstract

In recent years, significant interest in AI has resulted in many K-12 learning resources being developed for teachers and students. However, a consensus on which concepts and skills should be developed has yet to be reached. In addition, there is limited research on how these resources can be used by teachers. As AI/machine learning is not currently mandated in many CS curricula, supporting teachers is an ongoing challenge. Analysing resources currently available to teachers provides us with a starting point to consider how resources could be designed to better support teaching and learning in this area. In this paper, we surveyed the landscape of resources available and categorised 307 AI/ML teaching resources based on their core learning focus. We devised and employed a learning content analysis framework, SEAME, to determine how resources covered AI at multiple levels: (i) Social & Ethical; (ii) Application; (iii) Model, and; (iv) Engine. We found that most resources focused on the Application level (80%) with many covering the Model level (60%) and fewer at the Social & Ethical (40%) and Engine (27%) levels. We found little consensus about what to teach and how with many resources failing to specify target ages. Similarly, we found few examples of resources with professional development opportunities or appropriate lesson documentation, indicative of the challenges teachers face in teaching about AI. We propose that the SEAME framework provides an innovative starting point for teachers and researchers to review resources and consider what a progression of AI concepts and skills might look like that is comprehensive and simple to use. Likewise, it provides a common language for articulating the learning focus of resources. In order to support teachers encountering a new strand of CS, we suggest that further work in developing age-appropriate curricula and better documentation for AI resources is needed.

Topics & Concepts