关键词: 人机协同决策, 观点演化, Hegselmann-Krause模型, 人机信任, 机器推荐精度

Abstract: The rapid advancement of artificial intelligence (AI) and the increasing complexity of decision-making environments have revealed the limitations of traditional opinion dynamics models in characterizing human–machine collaborative decision-making processes. Meanwhile, the influence mechanisms of human–machine trust and algorithmic recommendation accuracy on individual decision behavior and collective opinion evolution remain insufficiently explored. To address these issues, this study extends the classical Hegselmann–Krause (HK) model by incorporating key factors such as individual social network structure, human–machine trust level, and recommendation accuracy, thereby constructing a dynamic model of human–machine collaborative decision-making. Within this framework, numerical simulations are conducted to systematically analyze the evolution patterns of group opinions under different trust structures and recommendation accuracies. Furthermore, the network topology and recommendation functions are extended to verify that the observed model effects arise from intrinsic mechanisms rather than external assumptions. Finally, empirical analyses based on real-world data demonstrate that the proposed model effectively captures the dynamic processes of individual behavioral evolution and collective consensus formation in human–AI collaborative contexts. The findings provide theoretical insights and methodological guidance for understanding AI-empowered human–machine collaboration and optimizing collective decision-making mechanisms.

Key words: Human–AI collaborative decision-making, Opinion evolution, Hegselmann–Krause model, Human–AI trust, Machine recommendation accuracy