Technological efficiency thresholds and sustainable development in G20 countries: the moderating role of artificial intelligence

Abstract

Despite the increasing interest in the roles of research and development (R&D), artificial intelligence (AI), and renewable energy (RE) in sustainable development, few studies have investigated their nonlinear, threshold-dependent relationships across G20 economies, highlighting the need for such an analysis. This article makes several unique contributions to the literature on environmental sustainability and the energy transition by examining AI as a transformative element. Using data from G20 nations over the period from 2000 to 2023, R&D is treated as a threshold variable that modulates the relationship between renewable energy and sustainable development. The analysis employs a panel smooth transition autoregression (PSTAR) model to capture nonlinear dynamics, allowing for regime-dependent impacts associated with technological maturity and AI diffusion. The findings identify three R&D efficiency thresholds (-1.057, -0.057, and 0.185), beyond which the effects of R&D, AI, and renewable energy on green growth change substantially. The transition functions illustrate smooth, incremental changes across regimes, confirming that countries progress through phases of technical growth gradually rather than experiencing abrupt shifts. In lower-efficiency regimes, the effects of R&D and renewable energy are limited or even detrimental. In contrast, in higher-efficiency regimes, where AI and innovation capabilities are more advanced, these effects become highly positive. Moreover, environmental degradation, as measured by CO₂ emissions, reduces the effectiveness of R&D, while AI amplifies it. Importantly, the robust system GMM estimates corroborate the PSTAR findings, reinforcing the non-linear nature of the relationship and providing empirical support for a Kuznets-type curve in green growth, where the benefits of R&D and technological innovation are realized beyond specific thresholds. These results provide concrete policy guidance by identifying critical thresholds and synergistic effects that should inform integrated innovation and sustainability strategies within the G20. Overall, the study contributes to achieving the Sustainable Development Goals (SDGs), particularly those related to renewable energy, innovation, and climate action.

Keywords: Artificial intelligence; Environmental sustainability; G20 countries; Non-linear modelling; R&D efficiency.