人工智能应用缓解新索洛悖论的机制研究 ——来自企业层面的经验证据

A Study on the Mechanism of AI Application Alleviating the New Solow Paradox: Empirical Evidence from the Enterprise Level

摘要:

通用人工智能应用正逐步在千行百业开展概念验证与场景探索，但其能否切实推动劳动生产率提升仍受质疑，因此人工智能应用能否缓解新索洛悖论成为社会关注的热点。本文基于2012—2023年A股上市公司数据，运用双重固定效应模型和双重机器学习模型，研究人工智能应用缓解新索洛悖论的机制。通过实证发现人工智能应用可以缓解新索洛悖论。机制分析表明，人工智能应用通过优化劳动力结构、减少组织冗余、强化组织能力来缓解新索洛悖论，且激励机制在其中发挥正向调节作用。异质性分析表明，非高技术行业比高技术行业人工智能的应用效果更为显著，东部和中部地区人工智能应用效果显著，而西部地区生产率效能提升受到约束。此外，在行业异质性中，人工智能的应用显著提升了制造业，服务业及电力、热力、燃气及水生产和供应业的生产率；而在农业、采矿业和建筑业中，其生产率提升效能受到约束。本文为企业应对新索洛悖论挑战、实现人工智能驱动企业全要素生产率跃升提供了实践启示与政策建议。

Abstract:

The application of general artificial intelligence is gradually undergoing concept verification and scenario exploration across various industries, yet whether it can effectively boost labor productivity remains questionable. Therefore, whether the application of artificial intelligence can mitigate the New Solow Paradox has become a topic of considerable concern. Based on data from A share listed companies spanning from 2012 to 2023, the two way fixed effects model and the double machine learning model were applied to examine the mechanism by which the application of artificial intelligence mitigates the New Solow Paradox. Empirical evidence reveals that artificial intelligence applications can mitigate the New Solow Paradox. Mechanism analysis indicates that artificial intelligence applications mitigate the New Solow Paradox by optimizing labor structure, reducing organizational redundancy, and strengthening organizational capability, with incentive mechanisms playing a positive moderating role in this process. Heterogeneity analysis reveals that the application effects of artificial intelligence are more pronounced in non-high-tech industries than in high-tech industries, and significantly enhance total factor productivity in the eastern and central regions, while their efficacy is constrained in the western region. Furthermore, in terms of sectoral heterogeneity, the application of artificial intelligence significantly improves total factor productivity in manufacturing, services, and the production and supply of electricity, heat, gas, and water, while its productivity enhancing effects are constrained in agriculture, mining, and construction. Practical implications and policy recommendations are offered for enterprises to cope with the challenges of the New Solow Paradox and achieve leapfrog growth of enterprise total factor productivity driven by artificial intelligence.