When proximal collaboration with AI hinders innovation: The moderating role of idea originality and reliance on AI

Although the importance of human–AI collaboration for organizational performance is increasingly acknowledged, its relationship with innovation remains fragmented. Based on tenets of Distributed Cognition theory, we predict that proximal human–AI collaboration impairs key innovation behaviors, namely idea selection and idea implementation. Additionally, we predict that idea originality mitigates the negative effect of proximal human–AI collaboration on idea selection, whereas humans' reliance on AI's input mitigates its negative effect on idea implementation. Results from an experimental study using a one-factorial (proximal human–AI collaboration: high vs. low) between-subjects design and a sample of 221 employees showed that proximal human–AI collaboration decreased idea selection, particularly when idea originality was low. The results further showed that proximal human–AI collaboration reduced idea implementation conditional on reliance on AI, with the negative effect weakening as reliance increased. Our findings advance understanding of the risks of human–AI collaboration for innovation by showing that proximal collaboration reduces idea selection and idea implementation, with effects shaped by idea originality and individuals' reliance on AI, respectively.