Norwegian University of Science and Technology, Norway
Studies show that creativity can be augmented by AI, but it is not clear how, or by what process, that can happen. This paper explores how organizational members use and engage with language models to augment their creativity. Our findings indicate that while language models can be prompted to generate ideas, the generation of ideas does not in itself augment creativity. Rather, AI-augmented creativity lies in refining ideas through iterative perspective taking which increases comprehensiveness and facilitates identification and selection of the most novel and useful ideas. Our model of AI-augmented creativity illustrates that while idea generation leads to idea access efficiency, idea development leads to idea comprehensiveness which enables the selection of creative (novel and useful) ideas. Our study contributes to creativity theory in three significant ways: First, our study suggests that using language models for creative tasks makes the creative process dynamic, incorporating both idea generation and development. Second, our study suggests that in addition to task-domain knowledge, creativity-relevant processes, and intrinsic motivation, generative AI knowledge is an important factor for AI-augmented creativity. Third, our study suggests that the value of language models is not only in providing breadth of ideas but also in enabling comprehensiveness of thought.
