Research on Iterative Diffusion Models of Generative AI Technology

Abstract

To explore the causes of the digital divide in the generative AI era and address the limitations of traditional innovation diffusion theory, this study adopts a dual framework of innovation diffusion and reflexivity theory. Using content analysis and indepth interviews, it focuses on DeepSeek from December 2024 to December 2025 to examine the iterative diffusion of generative AI.The study finds that generative AI diffusion features simultaneous technological iteration and adoption driven by audience rational cognition, breaking traditional diffusion laws. Consumers’ technological maturity expectations and social factors interact and reinforce each other as core barriers, forming the new digital divide mechanism.Replacing the traditional Sshaped curve model, this study constructs a stepwise upward model of iterative diffusion, clarifies the negative impacts of audience refusal at individual, technical and social levels, and answers three core questions about refusal reasons, laggard factors and related impacts. It confirms that the digital divide is shaped by technological cognition, maturity expectations and social factors.Finally, it proposes improving digital literacy, guiding rational technological expectations and fostering a positive communication environment to bridge the divide. This study enriches the two theories in the digital context, fills the userperspective gap in generative AI research, and provides a new theoretical model and perspective for digital technology innovation diffusion studies.