Teaching Social Innovation in the Age of AI

Authors

  • Mats Danielson

DOI:

https://doi.org/10.65343/erd.v2i1.65

Keywords:

design thinking, social innovation, artificial intelligence, concept generation, public sector innovation, project based learning, higher education, human AI collaboration, qualitative case study, Openlab Stockholm

Abstract

This paper examines how a contemporary conversational artificial intelligence performs in the concept generation phase of design thinking based social innovation education. The setting is Openlab Stockholm’s interdisciplinary master course Innovations for Societal Challenges, where student teams work with real challenges from the City of Stockholm and Region Stockholm. In the study, eight project reports from recent course iterations were selected, of which six could be processed technically. For each team, a three step protocol was applied. First, the AI chatbot (ChatGPT 5.1 Pro) read only the empathise and define sections of the student report, including field insights and the team’s reframed point of view, but not their own concepts. Second, the chatbot generated five solution directions, called concepts in the course, based solely on that material. Third, once the full report including the student concepts was opened by the author only, the two concept sets were compared along five criteria: social impact and relevance, innovativeness and insightfulness, feasibility and solidness as a foundation for continued work, agency and ownership including stakeholder alignment, and scalability, learning and robustness. Each set was given an indicative score on a five-point scale for every criterion and an overall average per case.

Across the six complete comparisons, the chatbot produced concept sets that were consistently competitive with the student teams, but typically scored slightly lower overall. The AI performed particularly well on tight alignment with user needs, clear conceptual structuring and proposals for modular, scalable engagement systems. The student teams, however, outperformed the AI on contextualisation in the Swedish public sector, institutional realism, stakeholder anchoring and empirically grounded feasibility, supported by prototyping, surveys and sponsor interaction. The study concludes that a current AI chatbot can match master students in creativity and user insight when given high quality research and problem framing, but cannot replace the situated, collective work of embedding concepts in real organisational, legal and political contexts. In design thinking education, its most appropriate role is that of a powerful early stage ideation partner rather than a substitute innovator.

Downloads

Published

2025-12-17

How to Cite

Danielson, M. (2025). Teaching Social Innovation in the Age of AI. Educational Research and Development, 2(1), pp. 21–42. https://doi.org/10.65343/erd.v2i1.65