Die Akzeptanz von KI-Anwendungen in Scrum-Teams

: Perspektiven von Product Ownern, Scrum Mastern und Developern

Abstract

Although generative systems — such as large language models and autonomous agents — are increasingly employed in software development, there is little empirical research into which factors determine their use in Scrum settings, or whether these drivers affect Product Owners, Scrum Masters, and Developers differently. The present Master’s thesis investigates how AI applications are adopted in Scrum teams, integrating established technology-acceptance models with the specific challenges of agile environments. To close this gap, nine semi-structured expert interviews were conducted — three practitioners from each of the three responsibilities — using an interview guide based on the UTAUT constructs. The transcripts were then analyzed via Mayring’s qualitative content analysis. In addition to the five deductively defined main categories (Performance Expectancy, Effort Expectancy, Social Influence, Facilitating Conditions, and Attitude Toward Using Technology), an additional main category — Data Protection & Transparency — and its subcategories emerged inductively. The empirical findings reveal that organizational and technical framework conditions — for example, targeted training programs and seamless tool integration — exert a strong influence on the acceptance of AI applications in Scrum teams. Other significant drivers include expected performance gains, perceived effort, social influence from colleagues and leadership, and personal attitudes toward the technology, which varies according to the user’s experience and the application scenario. The newly identified Data Protection & Transparency dimension proves to be an additional trust factor, particularly in sensitive contexts. Role-specific differences are evident: Product Owners place the greatest emphasis on framework conditions. Scrum Masters highlight social and affective factors, while Developers focus primarily on technical integration. Building on these insights, extensions to existing acceptance models are proposed: the addition of a trust/data-protection dimension, the accommodation of role-specific requirements, and the introduction of a contextual influence factor for agile settings. From a practical standpoint, this yields concrete recommendations — such as tailored training programs for each role, curated prompt libraries, and integrated solutions within existing systems. Thus, the thesis offers both theoretically grounded approaches for shaping AI-adoption strategies in agile teams and practical guidelines for the sustainable implementation of AI applications.

Date of Award	2025
Original language	German (Austria)
Awarding Institution	Johannes Kepler University Linz
Supervisor	Anita Vogl (Supervisor)