Abstract
Artificial intelligence (AI), particularly machine learning (ML), has become nearly ubiquitous in recent years. From social media to medical work, AI provides the infrastructure to many of our most used and critical IT systems. However, such systems are prone to unintended biases rooted in historical data, their inner workings are opaque, and we have seen intentional nefarious use in, e.g., democratic elections. This development calls for a democratization of AI: that its further development and use be made political and that public debate is fostered around it. Qualifying this necessitates that the public develops a critical understanding of what constitutes an AI system, how such systems are designed, and what potential impacts they might have. The recency of the proliferation of AI means that while some suggestions for AI curricula exist, the design space for concrete learning tools and activities is yet ill-defined. Most previous work focuses on developing children’s skills and competencies with regard to AI, and little work exists that explicitly aims to support critical understanding of and attitudes toward it.
Through a series of constructive design-research experiments aimed at sec- ondary school students, this dissertation explores the design space for learning tools and activities for computationally empowering ML education: ML education that engages learners in understanding ML to scaffolding critical reflection on its role in their own life, and in society around them. Based on six experimental designs and classroom interventions with these, the thesis makes three main contributions to HCI, child-computer interaction, and computing education.
First, by analyzing the experiments and discussing them in relation to the literature, I present six concrete design principles for AI learning tools and activities with computational empowerment as the goal. Further, to qualify how these principles interact in concrete designs, I discuss their tensions and synergies.
Second, I present the CEML framework: an approach to computationally empowering ML education based on a technological, material foundation that highlights the ethics and morality baked into technology and offers concrete ways of including this in computing education.
Finally, I reframe my original research program and return to HCI to present Remarkable AI: an approach to AI systems design that focuses on fostering agency in users and empowering them with regard to its role in their lives. In addition, I argue that designers and HCI researchers should consider remarkableness as a sensitizing concept when designing AI systems.
Together, these contributions are intended to support designers, researchers, and, not the least, educators in teaching and designing tools and activities that democratize AI and open it up for political discourse.
Through a series of constructive design-research experiments aimed at sec- ondary school students, this dissertation explores the design space for learning tools and activities for computationally empowering ML education: ML education that engages learners in understanding ML to scaffolding critical reflection on its role in their own life, and in society around them. Based on six experimental designs and classroom interventions with these, the thesis makes three main contributions to HCI, child-computer interaction, and computing education.
First, by analyzing the experiments and discussing them in relation to the literature, I present six concrete design principles for AI learning tools and activities with computational empowerment as the goal. Further, to qualify how these principles interact in concrete designs, I discuss their tensions and synergies.
Second, I present the CEML framework: an approach to computationally empowering ML education based on a technological, material foundation that highlights the ethics and morality baked into technology and offers concrete ways of including this in computing education.
Finally, I reframe my original research program and return to HCI to present Remarkable AI: an approach to AI systems design that focuses on fostering agency in users and empowering them with regard to its role in their lives. In addition, I argue that designers and HCI researchers should consider remarkableness as a sensitizing concept when designing AI systems.
Together, these contributions are intended to support designers, researchers, and, not the least, educators in teaching and designing tools and activities that democratize AI and open it up for political discourse.
| Originalsprog | Engelsk |
|---|
| Forlag | Århus Universitet |
|---|---|
| Antal sider | 236 |
| Status | Udgivet - apr. 2023 |