International Open Access Journal Platform

情境主导还是人格驱动？生成式人工智能信任与依赖对个体决策行为的影响

Situation-Driven or Personality-Driven? The Effects of Trust and Reliance on Generative Artificial Intelligence on Recommendation Adoption

姬昱辰, 姜诚, 廖思羽, 谭梦鸽

Psychology of China / 2026,8(6): 1018-1024 / 2026-06-30 14 11

View Download PDF

• Information:

  江苏理工学院教育学院应用心理学系，常州

• Keywords:
  Generative artificial intelligence; AI trust; AI reliance; Recommendation adoption; Big Five personality; Risk perception
  生成式人工智能; 人工智能信任; 人工智能依赖; 建议采纳; 大五人格; 风险感知
• Abstract: Generative artificial intelligence (AI) has been increasingly integrated into decision-making across various domains, making human–AI collaborative decision-making more prevalent. This study investigated the factors influencing recommendation adoption in generative AI-assisted decision-making. Fifty-three undergraduate students participated in the study. Participants first completed a self-developed Generative AI Trust–Reliance Scale and were classified into high- and low-trust groups. They then completed a 2 (risk level: high vs. low) × 2 (AI recommendation accuracy: high vs. low) within-subject experiment. Openness and neuroticism were additionally measured using the Big Five Personality Inventory. The results showed that both risk level and AI recommendation accuracy had significant main effects on recommendation adoption, and their interaction was also significant. Specifically, individuals were more sensitive to AI recommendation accuracy under high-risk conditions. In contrast, neither trait AI trust and reliance nor openness and neuroticism showed significant effects. These findings suggest that recommendation adoption in the early stage of human–AI collaboration is primarily driven by situational factors rather than stable individual differences, providing empirical evidence for trust calibration in human–AI interaction. 生成式人工智能（AI）技术快速渗透多领域决策场景，人机协同决策日趋普遍。为探讨生成式AI辅助决策情境下建议采纳行为的影响因素，本研究以53名大学生为被试，首先采用自编生成式AI信任—依赖量表测量个体AI信任与依赖水平，并依据量表得分进行高、低信任组划分；随后采用2（风险水平：高、低）×2（AI建议准确率：高、低）被试内实验设计，考察风险水平和AI建议准确率对建议采纳行为的影响，并结合大五人格量表考察开放性和神经质人格的作用。结果发现：（1）风险水平和AI建议准确率对建议采纳率均具有显著主效应；（2）风险水平与AI建议准确率存在显著交互作用，高风险情境下个体对AI建议准确率更加敏感；（3）AI信任水平以及开放性、神经质人格的作用均未达到显著水平。本研究揭示了生成式AI场景下建议采纳的情境主导机制，补充了个体差异的阴性实证证据，可为人机信任校准设计提供参考依据。
• DOI: 10.35534/pc.0806150 (registering DOI)
• Cite: 姬昱辰, 姜诚, 廖思羽, 谭梦鸽. (2026). 情境主导还是人格驱动？生成式人工智能信任与依赖对个体决策行为的影响. 中国心理学前沿, 8 (6), 1018-1024.