ISSN 1671-3710
CN 11-4766/R
主办：中国科学院心理研究所
出版：科学出版社

• •    

基于生成式AI的多模态商业内容最优区分性对网红种草营销效果的影响

洪傲然, 冯梓煜, 王永贵   

  1. 南京理工大学经济管理学院, 江苏 210094 中国
    浙江工商大学现代商贸中心，工商管理学院，中国智能管理研究院, 浙江 310018 中国
  • 收稿日期:2025-10-19 修回日期:2025-12-12 接受日期:2025-12-17
  • 基金资助:
    国家自然科学基金青年科学基金项目（C类）(72502108)

The influence of optimal distinctiveness of multi-modal sponsored content on influencer marketing effectiveness based on generative AI

Hong Aoran, Feng Ziyu, Wang Yonggui   

  1. School of Economics and Management, Nanjing University of Science and Technology 210094, China
    Modern Business Research Center, College of Business Administration (MBA) and c. Intelligent Management Institute of China, Zhejiang Gongshang University 310018, China
  • Received:2025-10-19 Revised:2025-12-12 Accepted:2025-12-17
  • Supported by:
    National Natural Science Foundation of China (NSFC) Youth Project (Category C)(72502108)

PDF (PC)

25

摘要： 生成式AI的发展显著提升了营销从业者的内容创作能力。但网红种草营销作为扩大内需、促进消费的重要引擎，正因内容创作同质化问题而陷入销售转化率低迷的现实困境。本研究提出，识别多模态商业内容的最优区分阈值，并应用生成式AI实现最优区分创作，是破解这一困局的关键路径。为此，本研究基于社交媒体可供性理论和最优区分理论，创新性地运用多模态机器学习、量化营销及行为实验等混合研究方法，致力于：（1）识别多模态商业内容的特征维度，测量多模态商业内容相似性；（2）厘清多模态商业内容相似性对网红种草营销效果的影响机制，进而确定相似性的最优区分阈值；（3）探究如何利用生成式AI创作具备最优区分性的多模态商业内容。研究成果将完善网红种草营销效果的理论分析框架，为破解实践难题提供切实可行的科学方案，不仅契合当下数字经济时代“AI+消费”商业生态发展的核心需求，更对扩大内需和提振居民消费具有显著的实践价值。

关键词: 网红种草营销, 消费者心理, 多模态分析, 生成式AI, 最优区分理论

Abstract: The rapid advancement of generative AI has significantly enhanced content creation capabilities for marketing practitioners. However, influencer marketing— a crucial engine for stimulating domestic demand and boosting consumption — now faces practical challenges of low sales conversion rates due to content homogenization. This study proposes that leveraging generative AI to create sponsored content adhering to the optimal distinctiveness principle may resolve this problem. Based on social media affordance theory and optimal distinctiveness theory, the research innovatively employs a mixed-methods approach integrating multimodal machine learning, quantitative marketing analysis, and behavioral experiments. It aims to: (1) identify feature dimensions of multimodal sponsored content and measure sponsored content similarity; (2) unveil the mechanisms of sponsored content similarity on influencer marketing effectiveness to identify optimal distinctiveness thresholds; (3) explore technical pathways for generating optimally distinctive multimodal sponsored content using generative AI. The findings will establish a comprehensive theoretical framework for analyzing influencer marketing effectiveness while providing actionable scientific solutions to practical challenges. The outcomes not only align with the core demands of developing an "AI+Consumption" ecosystem in the digital economy era but also demonstrate significant practical significance for expanding domestic demand and revitalizing consumer markets.

Key words: influencer marketing, consumer psychology, multi-modal analysis, generative AI, optimal distinctiveness theory