Towards Human-Centered AI-Based Tutoring in Higher Education: A Requirements Analysis of Student and Instructor Perspectives
Fulltext URI
Document type
Additional Information
Date
relationships.isAuthorOf
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Generative artificial intelligence (GenAI) tools have become part of students’ academic practices in higher education, where general-purpose systems are used for activities such as information seeking, text production, summarization, and exam preparation. However, such systems are often insufficiently aligned with course-specific content and learning objectives and provide limited opportunities for instructors to control the knowledge sources underlying system responses. This study presents a human-centered requirements analysis for generative AI-supported tutoring systems in higher education, focusing on the perspectives of students and university instructors. Semi-structured interviews were conducted with university instructors from different academic disciplines (N=6), and an online survey was completed by students (N=42). Students considered explanations of complex concepts, practice activities, exam preparation, summaries, and individualized feedback particularly relevant. They also emphasized transparency, instructor guidance, and reliable system responses. Instructors highlighted integration into existing learning management systems, control over the materials used as the system’s knowledge base, alignment with defined learning objectives, and transparent communication of system limitations. Overall, the findings suggest that human-centered AI tutoring in higher education requires balancing automated, low-threshold learning support with human control over pedagogically relevant system conditions. Central requirements include course-specific learning support, instructor-governed knowledge sources, transparent system behavior, reliability, and institutional integration.
