Abstract
Type of the article: Research ArticleAI is widely regarded by the IMF and the World Bank as a catalyst for growth. AI should be understood as a multidimensional socio-technical system embedded across institutions, industries, and society. Its economic contribution depends on which pillars of the national AI system expand (e.g., R&D capacity, infrastructure, governance, or social acceptance). For this reason, the seven pillars of AI development are measured by the AI Vibrancy subindices, which help avoid reliance on a single composite indicator that may conceal offsetting effects. This study examines how different pillars of the national AI ecosystem shape the architecture of the knowledge economy and its economic outcomes by estimating heterogeneous within-country associations between GDP per capita and seven AI ecosystem pillars, operationalized through AI Vibrancy subindices, using a balanced panel of 36 countries with complete data over the period 2020–2023. Fixed- and random-effects models are estimated using heteroskedasticity-robust and Driscoll-Kraay standard errors. The results indicate that, within countries over time, the R&D (β = –5.676, p < 0.001) and Infrastructure (β = –16.306, p < 0.001) subindices have strong and statistically significant negative associations with GDP per capita, while Public Opinion shows an adverse effect that is significant at the 5% level under heteroskedasticity-robust inference (β = –9.126, p = 0.040) and marginally significant under Driscoll-Kraay inference (p = 0.054). Responsible AI exhibits a marginally positive association (β = 5.773, p = 0.065) in the Driscoll-Kraay specification, whereas Economy, Education, and Policy & Government show no significant within-country effects.