The Influence of AI Algorithm-aided Decision on the Diffusion of Industrial Chain Integration Innovation Decision

Abstract

Algorithm-assisted decision-making is a crucial approach to enhance decision-making efficiency, scientific rigor, and objectivity. This study constructs a dual-layer coupled network model integrating evolutionary game theory and deep reinforcement learning to investigate the impact of algorithmic intervention on innovation decision diffusion.Simulation results demonstrate: (1) Embedding algorithm-assisted decision-making significantly accelerates innovation diffusion speed and stability, yet exhibits a dual-threshold effect in algorithm weight allocation. Additionally, diffusion dynamics are influenced by upstream-downstream integration innovation ratios, exploration rates, network scale, and initial diffusion proportions. (2) When initial algorithm weights approach thresholds, decision diffusion demonstrates strong adaptability to integration ratios, strategy update sensitivity, and initial diffusion proportions, whereas high exploration rates and undersized networks markedly suppress diffusion velocity. Low algorithm weights hinder integration innovation between upstream and downstream enterprises in driving diffusion. (3) When initial algorithm weights exceed thresholds, innovation diffusion becomes constrained, yet efficiency can be partially restored by increasing initial integration ratios and network scale. These findings provide theoretical foundations for optimizing enterprise innovation decision processes and guiding governmental policies on algorithmic governance and industrial chain coordination.