S1-STEM3-1 - Machine learning for Middle schoolers: Children as designers of ML Apps
1. Innovative Practice Full PaperTapani Toivonen1, Ilkka Jormanainen1, Matti Tedre1, Juho Kahila1, Teemu Valtonen1
1 University of Eastern Finland
Introduction
During recent years, a major technological shift has triggered discussions about the need to amend computing education at all education levels. Traditional, rule-based automation has been joined by machine learning (ML), which, when provided with enough computing power and data, has enabled new classes of jobs to be automated, and thus expedited automation in the society, workplace, and in people’s everyday lives. Yet, this development and its societal effects have been given minor attention in education, which mainly focuses on rule-based programming. Accordingly, there are few shared practices on how to introduce ML at the K-12 level in a manner that provides students the opportunities to develop the skills and mindset needed in data-driven society.
This study is a part of a multidisciplinary, design-based research project that aims to develop and study pedagogical models and tools for integrating ML topics into education. The design of the present study followed the currently popular research practice of including children as contributing members and meaning-makers in cutting-edge practices of research and development work. The pedagogical rationale behind that participatory research approach is to empower children to make a difference in the technological world around them as well as to enable children’s voices, views, ideas, and experiences to be heard when introducing machine learning to K-12 education.
This paper presents a pedagogical framework for supporting middle schoolers to become co-designers and makers of their own machine learning applications. We conducted a series of co-design workshops with primary students (N=34), where they explored and co-designed their own machine learning driven applications. Data consists of pre-and post test, students’ design ideas and co-designed applications, and structured group interviews organized at the end of the ML project. This paper analyzes the ML co-design process and ML applications that students designed for solving meaningful problems in their everyday life.
The qualitative content analysis revealed how hands-on exploration with ML-based technologies supported students to develop various kinds of design ideas that harnessed facial recognition, gesture recognition, and sound recognition for solving real-life problems. The results of the study further indicated that co-designing ML applications provided a promising entry point for students to develop their conceptual understanding concerning ML and apply it in the practice of computational design. What is more, working with ML also supported students to better understand the underlying principles of the ML systems that they are exposed to in their everyday lives. We conclude with a discussion on how to expand the ML design possibilities for students while also offering them new ways to make sense about the data-driven world that they live in.