In many organizations, integrating artificial intelligence (AI) into creative innovation processes remains insufficiently structured, despite the widespread availability of AI technologies across nearly all industries. Particularly within the context of service design, there is a lack of systematic understanding regarding how AI can effectively support the individual phases of the Design Thinking Process (DTP). Against this backdrop, it is imperative to reflect on current practice and contribute to closing this gap. Starting from the observation that existing studies on the use of AI in the DTP yield fragmented and sometimes contradictory findings, this thesis formulates two central research questions: first, how AI can enhance the five phases of Empathize, Define, Ideate, Prototype, and Test in service design; and second, whether AI functions primarily as a mere tool or as an active team member. To answer these questions, a qualitative grounded-theory approach is employed, combining a literature review with 15 semi-structured expert interviews. The analytical framework is based on open, axial, and selective coding of the transcribed and validated interview data. Theoretical sampling ensures that diverse perspectives from various countries and roles within service design are taken into account. The results demonstrate that AI offers substantial benefits across the first four phases: it accelerates data synthesis in the Empathize phase, supports problem definition through automated clustering techniques, expands the ideation space in the Ideate phase via generative models, and enables rapid prototype creation. At the same time, experts emphasize that genuine user insight, critical reflection, and human judgment remain indispensable. AI’s role operates along a continuum from “tool” to “assistant” to “sparring partner,” depending on individual expertise and project-specific needs. Fully autonomous applications currently appear only marginally practicable. Finally, a theoretical framework is proposed that integrates phase-specific gains, restrictions, and moderating factors (such as human intervention and expertise). Future research should aim to empirically validate this framework in real-world projects to assess its generalizability and long-term impact.
Christian Emil Höfler (Thu,) studied this question.