227. Cognitively appropriate and readily accessible computing education for young learners Book (2022) Dietz, Griffin Gabrielle, author Gweon, Hyowon, degree supervisor. thesis advisor Landay, James A., 1967-, degree supervisor. thesis advisor Piech, Chris (Christopher), degree committee member. thesis advisor Stanford University. Computer Science Department, contributor more In recent years, we have seen a growing push for computing education for all children. Unfortunately, computational thinking, programming, and artificial intelligence (AI) education reach only a fraction of young learners, in part because relevant educational tools often require expensive hardware or do not align with the cognitive skills and abilities (e.g., literacy) of young users. To fill these gaps we must both understand the developmental capacities of children to reason about and engage with computing concepts and design educational platforms that support age-appropriate learning. This dissertation describes research to that end across two domains of computing education—computational thinking and artificial intelligence. Specifically, this work 1) studies the ways that children intuitively think about everyday technology and their abilities to reason in a computational way, 2) understands the non-pedagogical needs of learners aiming to engage with this material, and 3) applies this knowledge to the design and development of accessible, approachable, and engaging systems for computing education. In the space of computational thinking, I show that, contrary to early work suggesting that elementary school children cannot engage with abstract content, children of this age can and do reason abstractly, and that appropriately scaffolded and taught computational practices (e.g., abstraction and decomposition) should be of greater focus in early computing education. Through an experimental behavioral lab study, I demonstrate that children have the capacity to engage in spatial decomposition tasks by early elementary years. I also report on a formative investigation that demonstrates how we can better support these learners through more accessible, approachable, and engaging tools. I then describe two software systems informed by these findings: StoryCoder, a voice-driven application to introduce key computing concepts to children, and Visual StoryCoder, a multimodal extension to that system that scaffolds computational practices as well. I then turn to artificial intelligence, a still-nascent subarea within computing education. While much of the effort in this space has focused on helping children learn about and use supervised learning classifiers, this type of machine learning is neither reflective of the kinds of AI technology that children regularly interact with nor easily-integrated into coding-centric computing curriculum. In this section, I describe an experimental behavioral study that demonstrates how children ages 3--8 might extend their cognitive capacities to reason about humans in order to reason about virtual assistants and conversational AI, one of the most widespread forms of AI today. This research demonstrates a need for education around AI and machine learning in K-12 education, and subsequent needfinding work emphasizes the importance of integrating that learning with learning about programming. Building on these findings, I then describe ARtonomous, a system for middle school reinforcement learning education in a virtual robotics context. Through this work, I demonstrate how, armed with an understanding of children's competencies and needs, we can design and develop cognitively appropriate and readily accessible computing education platforms for young learners more Available online See record