生成式人工智能与数字产品出口——来自生成式人工智能服务备案的证据

Abstract

As a core driving force of the new wave of technological revolution, generative artificial intelligence (AI) is profoundly reshaping the global trade landscape. This paper uses generative AI service registration data, regional export data, and firm-level bill-of-lading data from the S&P Panjiva database, combined with the BERT-wwm-ext Chinese model, to examine the impact of generative AI on digital product exports at both macro-regional and micro-firm levels. The study finds that generative AI significantly promotes China's digital product exports. Mechanism analysis shows that generative AI works through three channels: optimizing cross-border search and matching efficiency based on interactive capabilities; expanding export product variety based on generative capabilities; and improving export product quality based on 'general-purpose' capabilities. Heterogeneity analysis indicates that, at the technological level, generative AI applied to commercial scenarios such as e-commerce and product design, as well as those providing web-based access, has a more pronounced effect on digital product exports. At the product level, generative AI has a greater impact on the export of non-standardized and interactive digital products. At the regional level, the promotional effect is more significant in regions with higher e-commerce development, stronger computing power, and AI pilot zones. This paper provides empirical evidence for understanding the impact of generative AI on international trade and offers references for China's promotion of generative AI development.