Human agents, generative AI, and innovation: A formal model of hybrid creative process

Generative AI (GenAI) has rapidly emerged as a revolutionary technology that enables new ways to generate and recombine knowledge. Despite its significant potential, research on GenAI's role in enhancing creativity and innovation is still in its early stages. The present work advances this emerging field by focusing on the human-GenAI dyad. Specifically, we propose a formal model of hybrid creative process designed to maximize the synergistic potential of the human-GenAI interaction. Drawing on machine learning literature, we conceptualize GenAI as a superposition of latent entities. Through formal argumentation, we demonstrate that optimal creative outcomes arise when human agents actively select the most appropriate entity from the complete spectrum of potential alternatives for the problem at hand. Finally, we outline the ideal iterative process required to asymptotically converge toward these optimal entities. Beyond its practical utility for managers, our model provides new insights into human-GenAI mutual augmentation, the nature of creativity, and the skills and cognitive properties involved. • We conceptualize GenAI as a superposition of latent entities. • We show that human-GenAI creativity maximization requires selecting the optimal entity. • We introduce Bayesian optimization for optimal GenAI entity elicitation. • We leverage problem modularization as a means of human-GenAI capability integration. • We identify new creativity-relevant skills in the age of AI.