Designing Conversational Learning Companions from a Value-centered Perspective

With the increasing application of conversational AI in education, the role of conversational Learning Companions in enhancing students' learning process has gained significant attention worldwide. Technological advancements in artificial intelligence have already transformed the way we interact with information technology, from chatbots assisting in customer service to voicebots supporting us in daily life. Conversational agents leverage human patterns and design elements to engage with us, employing natural language and human-like appearances. In this dissertation, we explore the potential value of bonding Learning Companions, specifically focusing on the value-in-interaction, the value-in-use perceived by the human interactors and the co-created value between students and conversational AI in facilitating learning via positive and recurring interactions. Through a rigorous application of the Design Science Research Paradigm in several iterations, this research project incorporates multiple cycles of interviews, interdisciplinary workshops, prototype design and mixed method evaluations complemented by real-world testing with (German and international) universities or further education students from diverse disciplines. The research results encompass design knowledge on various abstraction levels (i.e., design principles, design features, situational artifacts) centered around different value layers that are crucial for nurturing effective interactions between students and their artificial conversation partners. Moreover, we approach the appendant Interactions-as-a-Service within the educational domain, offering practical insights from prototype testing in real learning scenarios and presenting an unconventional, value-driven theoretical perspective on designing Learning Companions as valuable dialogue systems. More precisely, this dissertation comprises a series of papers that comprehensively address various aspects of the underlying value of a Learning Companion’s design. Paper 1 introduces a teaching framework for Design Science Research, actively involving higher education students as co-creators in the research project. The subsequent Papers 2 to 4 delve into the fundamental concept of value-in-interaction in Learning Companions, with each paper focusing on a specific value layer: Value in Relatedness (Paper 2) to emphasize on power distance influencing relatedness between different cultural user groups, Matching Value (Paper 3) to highlight the potential of individualization in learning, and Service Value (Paper 4) in showcasing the Interaction-as-a-Service taking the example of bonding and time managing Learning Companions. To address ethical considerations, Paper 5 explores design features aimed at ensuring ethically responsible design practices, mitigating the potential for value co-destruction arising from e.g., biased data exchange between the Learning Companion and interacting students. Synthesizing the research findings, Paper 6 contributes value-centered design knowledge for Learning Companions across different levels of abstraction. Building upon this knowledge base, Paper 7 presents a refined Learning Companion and examines its impact on improving students’ time management, motivation to study and ultimately learning performance in a longitudinal study throughout a semester. Finally, Paper 8 investigates the concept of value co-creation between humans and conversational AI by integrating the design knowledge into an ideation companion for creating business models in entrepreneurship education. Conclusively, this dissertation contributes new insights into the value-driven design of conversational Learning Companions. It proves their positive impact on learning facilitation, but also highlights challenges in their ethically responsible design and technology adoption and the great need for adaptation to different contexts and user groups.