Meet, Greet or Tweet: Empowering Employees with Role-Optimized AI Assistance
(A2025-126091)

AUTHORS

Claus Hegmann-Napp, University of Hamburg; Tijmen Jansen, University of Hamburg; Mark Heitmann, University of Hamburg

ABSTRACT

This paper investigates the transformative potential of fine-tuned large language models (LLMs) in product design and innovation, focusing on their role in analyzing customer reviews and fostering human-AI collaboration. By leveraging a dataset of Amazon reviews, the research fine-tuned an LLM to synthesize customer insights and generate innovative product ideas. The study includes three empirical investigations: the comparative quality of ideas from fine-tuned LLMs, base LLMs, and human participants, the effectiveness of presenting AI as expert collaborators in enhancing human creativity, and the impact of different communication modalities (text, audio, and video) on the phases of idea creation. Results demonstrate the superior ability of fine-tuned LLMs in addressing consumer issues, enhancing purchase intent, and supporting human ideation processes. Participants expressed positive perceptions of AI collaboration, particularly in structured and innovative workflows. This research advances the integration of AI in design and innovation, emphasizing the synergy of human expertise and AI tools to redefine creativity and productivity.