Artificial Intelligence Investments and Environmental Quality: Evidence from Carbon Emissions, Ecological Footprint, and Sustainable Development Performance
Author :
Abstract
This study examines the impact of artificial intelligence investments on environmental sustainability by simultaneously considering three complementary environmental indicators: carbon emissions, ecological footprint, and sustainable development scores. The analysis employs a panel dataset covering the 19 countries with the highest levels of artificial intelligence investments over the 2012–2023 period. Long-run relationships among the variables are confirmed using the Westerlund-Edgerton Bootstrap cointegration test. Long-run coefficients are estimated using the Panel Corrected Standard Errors (PCSE), Seemingly Unrelated Regression (SUR), and Driscoll-Kraay estimators, while causal relationships are examined through the Dumitrescu-Hurlin panel causality test. The empirical findings reveal that artificial intelligence investments significantly increase carbon emissions and the ecological footprint, while significantly reducing sustainable development scores. Moreover, the results demonstrate that the environmental effects of artificial intelligence are not unidirectional and cannot be evaluated independently of countries' energy infrastructure and environmental policies. The findings are expected to provide important empirical evidence for policymakers, public authorities, and decision-makers responsible for designing environmental, energy, and digital transformation policies.
Keywords
Abstract
This study examines the impact of artificial intelligence investments on environmental sustainability by simultaneously considering three complementary environmental indicators: carbon emissions, ecological footprint, and sustainable development scores. The analysis employs a panel dataset covering the 19 countries with the highest levels of artificial intelligence investments over the 2012–2023 period. Long-run relationships among the variables are confirmed using the Westerlund-Edgerton Bootstrap cointegration test. Long-run coefficients are estimated using the Panel Corrected Standard Errors (PCSE), Seemingly Unrelated Regression (SUR), and Driscoll-Kraay estimators, while causal relationships are examined through the Dumitrescu-Hurlin panel causality test. The empirical findings reveal that artificial intelligence investments significantly increase carbon emissions and the ecological footprint, while significantly reducing sustainable development scores. Moreover, the results demonstrate that the environmental effects of artificial intelligence are not unidirectional and cannot be evaluated independently of countries' energy infrastructure and environmental policies. The findings are expected to provide important empirical evidence for policymakers, public authorities, and decision-makers responsible for designing environmental, energy, and digital transformation policies.
