Artificial Intelligence and the Acceleration of Innovation: Evidence from Advanced and Emerging Economies

Artificial intelligence has emerged as a central driver of digital transformation, with the potential to reshape the process of innovation across economies. In this context, the paper analyses the impact of AI-driven digitalization on innovation outcomes, using patent applications as a proxy, for a panel of countries comprising the G7 and G20 over the period 2004–2024. The empirical approach relies on static panel data estimations, including pooled OLS, fixed-effects, and random-effects models, complemented by alternative specifications controlling for country-specific and time-specific heterogeneity. In addition, Method-of-Moments Quantile Regression (MMQR) models are employed as a robustness check to examine whether the AI-innovation relationship differs across the patent distribution. The main findings reveal that the relationship between AI and innovation is heterogeneous, non-linear, and sensitive to model specification. For advanced economies (G7), the results provide stronger evidence of an inverted U-shaped relationship, whereby AI initially stimulates innovation by enhancing knowledge recombination and research efficiency, but exhibits diminishing marginal returns at higher levels of intensity. In contrast, for the broader and more heterogeneous G20 sample, the mean-based results are weaker and less robust, while the MMQR results suggest that AI becomes relevant mainly among higher-patenting observations. These results suggest that AI does not operate as an autonomous driver of innovation, but rather as a technology whose impact depends critically on absorptive capacity, complementary assets, institutional quality, and the position of countries within the innovation distribution.