Drivers of carbon emissions in the era of artificial intelligence: case of Japan

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

🔬研究者:Provides empirical evidence of a two-way causal link between AI and carbon emissions using ARDL models, contributing to the AI-environment debate.

🏛政策担当者:Recommends an integrated approach linking AI, energy, and environmental policies, and accelerating the shift to renewable energy and green AI technologies.

This research is driven by the absence of a unified consensus regarding the relationship between artificial intelligence, energy consumption, and economic growth and their impact on CO2 emissions. Not clear whether AI increases or decreases CO2. The novelty – identification of the two-way causal link between the implementation of AI and carbon emissions, a dynamic not previously confirmed in the literature. The originality – the model tested on a scale of Japan as developed country which still around 90% depending on fossil fuels. Data period: 1995–2024. An ARDL-based econometric approach to analyze the long-term and short-term impacts and several diagnostics to improve the precision and reliability of the study results. Tests used: the Breusch-Godfrey Serial Correlation, Ramsey RESET, Breusch-Pagan-Godfrey, CUSUMSQ and CUSUM. The study outputs reveal that in Japan, AI and energy consumption are associated with an increase in carbon emissions, while exports – with decrease in emission levels. In developed economies governments recommended to lower CO2 emissions by speeding up the shift toward renewable energy sources and investing in environmentally friendly AI technologies. Policymakers should adopt an integrated approach that links AI, energy, environmental, and economic policies, supported by regulatory reforms to promote sustainability and achieve carbon neutrality. First published online 8 June 2026

• crossref https://doi.org/10.3846/tede.2026.26996first seen 2026-06-16 05:39:04