Children’s Intuitive Theory of AI – an Active Training Study (Affiliated)

Many children today have never experienced life without Artificial Intelligence (AI). They are the first generation of truly digital natives. However, without understanding the mechanisms behind the technologies they are using, they cannot be fully competent users: confident and trusting, yet also critical in evaluating the strengths and limitations of AI. In this project, we aim to capitalize on preschool children’s natural curiosity-driven learn in order to teach them about AI. The goal is to provide children with an intuitive theory of AI that will help guide their AI interactions, similar to how a theory of mind guides their interactions with other humans.

It all starts with a robot appearing in the preschool, its goals, desires, and thoughts unclear to children. Through a series of interactive games, children learn to communicate with the robot and understand its behavior. Children are taught various information processing approaches (decision trees, heuristics, probabilities, causality, nearest neighbor classifications, supervised and unsupervised learning) through digital or analog games and are given the opportunity to apply their knowledge during interactions with the robot. Each lesson is performed collaboratively by a group of children, guided by their teacher to interact with and learn the embodied AI system. In control groups, children learn programming (Scratch Jr) or digital tool use (e.g. taking pictures, making movies, playing games).

Learning how to interact with the robot and getting an insight into the processes that guide AI systems should lead to enhanced problem-solving skills, AI self-efficacy (confidence in ones own ability to interact with AI systems), trust in AI (the belief that interactions will be positive) and critical thinking (understanding the limitations of AI). An increased AI competence early in life should also reduce gender stereotypes and decrease the gap in later interest and aspiration toward careers in science and technology.

Principal Investigator: Gustaf Gredebäck, Professor, Uppsala University
Co-Principal Investigator: Christine Fawcett, Associate Professor, Uppsala University
Marcus Lindskog, Associate Professor, Uppsala University
Ginevra Castellano, Professor, Uppsala University

19 October 2021 until 19 October 2021