科研管理

科研管理 ›› 2025, Vol. 46 ›› Issue (10): 72-81.DOI: 10.19571/j.cnki.1000-2995.2025.10.008

• 论文 • 上一篇    下一篇

人工智能驱动企业创新的生命周期异质性研究

余江1,2,李婉晴1,2,陈凤1,2,卢燃1,2   

  1. 1.中国科学院科技战略咨询研究院, 北京100190；
    2.中国科学院大学公共政策与管理学院, 北京100049
  • 收稿日期:2024-04-20 修回日期:2025-04-30 接受日期:2025-05-03 出版日期:2025-10-20 发布日期:2025-10-14
  • 通讯作者: 李婉晴
  • 基金资助:
    国家自然科学基金项目:“数字技术创新机制、突破路径和政策体系研究”（72334007，2024.01—2028.12）；教育部人文社会科学重点研究基地重大项目:“科技强国与高水平科技自立自强的理论逻辑与突破路径”(25JJD630001，2025.01—2026.12)。

Research on the lifecycle heterogeneity of AI-driven corporate innovation

Yu Jiang1,2, Li Wanqing1,2, Chen Feng1,2, Lu Ran1,2   

  1. 1. Institutes of Science and Development, Chinese Academy of Sciences, Beijing 100190, China; 
    2. School of Public Policy and Management, University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2024-04-20 Revised:2025-04-30 Accepted:2025-05-03 Online:2025-10-20 Published:2025-10-14

摘要：    人工智能技术已成为深化企业创新理论研究的重要工具，其应用也日益嵌入企业创新实践，对创新效率产生深刻影响。为了系统探讨人工智能技术对企业创新的影响机制，本研究基于企业生命周期视角，选取2016至2021年间1103家中国制造业高新技术上市企业数据，综合运用决策树和多种机器学习模型，探究人工智能技术在不同生命周期阶段对企业创新能力的异质性影响。研究发现：（1）多数机器学习模型对企业创新绩效（专利授权数）的预测效果显著优于传统线性回归模型，体现了机器学习在揭示变量间非线性关系方面的优势；（2）人工智能能力和专利申请数量的提升显著促进了成熟期企业的创新能力，但对成长期和衰退期企业的创新影响相对有限；（3）决策树分析表明，成长期和成熟期企业的创新更依赖于人工智能专利布局，而衰退期企业创新则更多依赖员工数量和技术人员规模。综合来看，人工智能技术总体上与企业创新绩效呈正相关，但在企业生命周期的不同阶段，这种关系存在显著的异质性。本研究不仅丰富了企业生命周期理论的内涵，也为有关企业针对性地制定阶段性人工智能战略提供了实践指导。

关键词: 人工智能, 企业创新, 企业生命周期, 决策树, 机器学习

Abstract:  Artificial intelligence (AI) technology has become an essential tool for deepening the theoretical research on corporate innovation, thus increasingly embedding itself into enterprise innovation practices and significantly impacting innovation efficiency. Grounded in a corporate lifecycle perspective, this study leveraged the data from 1,103 Chinese hightech manufacturing enterprises listed between 2016 and 2021, by employing the decision tree analysis and multiple machine learning methods to examine AI′s heterogeneous effects on enterprise innovation across different lifecycle stages. The findings are as follows: (1) Most machine learning models demonstrate superior predictive performance in forecasting firms′ innovation performance (patent authorizations) compared to traditional linear regression models, highlighting machine learning′s capability in capturing nonlinear relationships among variables. (2) Enhancing AI capability and increasing AI patent filings significantly boost innovation capacity in maturestage firms, while the innovation performance of growth and declinestage firms depends less on AI technology. (3) The decision tree analysis further indicated that innovation in growth and maturity stages primarily relies on AI patent activities, whereas innovation in declinestage firms depends more heavily on the size of employees and technical personnel. Overall, while AI technology positively correlates with enterprise innovation performance, this relationship varies significantly across corporate lifecycle stages. This study will extend the lifecycle theory and provide practical guidance for enterprises on strategically deploying AI technology according to their developmental stages.
Keywords:

Key words: artificial intelligence, corporate innovation, corporate lifecycle, decision tree, machine learning