Designing Interactions Between Students and a Virtual Agent in Problem-Solving Contexts#
- Oliver Paszkiewicz
- Master of Applied IT
Abstract
This study explores the design and evaluation of interactions between undergraduate students and a virtual agent aimed at supporting literature review tasks in educational settings. Grounded in the principles of human-agent collaboration and interaction design, the project investigates how virtual agents can complement students’ cognitive efforts and enhance problem-solving capabilities. Using the Wizard of Oz methodology, this research simulates student-agent interactions to identify preferences for structured guidance, feedback mechanisms, and adaptive interface designs. Findings emphasize the value of Localization Feedback and Direct Feedback, which provide actionable resources and clear explanations. A comparative usability analysis of movable widget and fixed sidebar interface designs reveals the advantages of flexibility and adaptability in enhancing user satisfaction and engagement. The results contribute to advancing human- AI collaboration in education, providing insights into optimizing virtual agent design for collaborative learning. Limitations and implications for future research are also discussed, highlighting the need for scalable implementations and expanded participant samples.