Research on the Mechanism of Artificial Intelligence-Driven Government Innovation Services to Enhance Manufacturing Innovation Capabilities

The rapid development of artificial intelligence has injected new momentum into manufacturing enterprises to enhance their independent innovation capabilities and strengthen their competitive advantages. Manufacturing enterprises should actively adapt to this development trend, seize the driving force provided by technological development, and enable Chinese manufacturing to form new advantages in global market competition. Therefore, this paper comprehensively reviews scholars' research on the artificial intelligence transformation of manufacturing and manufacturing innovation capabilities, proposes the theoretical foundation of innovation-driven and endogenous growth, and analyzes the current level of innovation capabilities of manufacturing enterprises under artificial intelligence technology. Using a two-way fixed effects model, we constructed a measurement framework and proposed relevant research hypotheses, designing a basic regression and mechanism testing model to analyze each research hypothesis individually. The findings indicate that the regression coefficient for the impact on the number of R&D personnel is 1.648, which is significantly positive at the 1% level. The application of artificial intelligence can increase the number of R&D personnel in an enterprise and promote employee participation in process innovation, thereby validating Hypothesis 1. Mediation effect tests were conducted simultaneously for industry type, innovation strategy, and government subsidies. When the explained variable was corporate innovation performance, the regression coefficients for industry type, innovation strategy, and government subsidies were 0.186, 0.197, and 0.185, respectively, all of which were significantly positive at the 5% level. This indicates that improvements in industry type, innovation strategy, and government subsidies can positively influence corporate innovation performance, thereby validating Hypotheses 2, 3, and 4.