Work

Pedagogical agents communicating and scaffolding student learning

Year: 2025
Type: dissertation
Abstract: Pedagogical agents are conversational learning technologies that interact with students through verbal (written or spoken) or non-verbal communication. Pedagogical agents can act in various roles in digital and virtual environments or be embodied like a robot. While they have facilitated student learning over two decades, their communication abilities have not been the primary focus of research. With the emergence of generative AI (GenAI) and large language models (LLMs), which are likely to enhance their communication capabilities, studying pedagogical agent communication is particularly timely. This study aims to contribute to the comprehensive body of knowledge on how pedagogical agents should communicate with students. This study’s contributions have been achieved through two systematic literature reviews, interviewing teachers and students, and conducting a study with high school students using communicative AI (Copilot) in their mathematics studies. Both qualitative and quantitative research methods were employed, depending on the study design, to provide the best approach for addressing the research problem. The dissertation’s contribution is twofold. First, it provides knowledge on how to develop appropriate and effective pedagogical agent communication. Second, it explores the communication between humans and machines by using the theoretical lens of human–machine communication. The study results show that positive, relational and affirmative pedagogical agent communication, which induces positive feelings, is the most effective in scaffolding student learning. In addition, when teachers and students outlined their preferred pedagogical agent communication, they applied human-to-human communication scripts. This indicates that human communication theories could be used as the baseline for developing pedagogical agent communication. When students used Copilot in their mathematics studies, perceived usefulness impacted their intention to use (ITU) GenAI, and perceived enjoyment (PE) impacted both perceived usefulness (PU) and ease of use (PE) statistically significantly. Also shown was that, despite their potential, these agents have been underutilized in serious games for autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD).
Author: Sikström, Pieta-Anniina
Language: Finnish
Open Access status: green
OpenAlex ID: W6995633778
Cites: 173
Cited by: 1
Related to: 10