Artificial intelligence (AI) is increasingly embedded in power grids, industrial operations and environmental monitoring, yet its sustainability benefits are conditional. AI can improve forecasting, renewable-energy integration and resource allocation, but it depends on reliable electricity, computing infrastructure, data, skilled labor and regulatory capacity—assets that are unevenly distributed. Expanding AI industries may raise electricity and mineral demand, accelerate hardware replacement and widen digital divides. These tensions matter for sustainable energy transitions, where efficiency gains can coexist with technological dependence. Given these challenges, in-depth research is needed on how AI diffusion affects different United Nations Sustainable Development Goals (SDGs) outcomes, how these effects vary by development level, and whether AI can reconcile efficiency and equity in sustainable energy transitions.
The study was conducted by a research team including first author Meiling Wu of Chengdu University of Technology, corresponding author Wei Liu of Nanchang University, and collaborator Zhengyang Zhang of Tohoku University, Japan. It was published (DOI: 10.1016/j.ese.2026.100784) online on 29 October 2026 in Environmental Science and Ecotechnology. Using double machine learning (DML), mechanism analysis and scenario projections, the team linked national AI development to seven SDGs across 69 countries from 2000 to 2023 and projected pathways to 2100. It identifies an efficiency–transformation gap: AI reliably supports education, clean energy and industrial innovation, but not systemic climate action or responsible consumption.
The study constructed a multidimensional AI development index covering patents, industrial robots, scientific publications, digital infrastructure, research and development (R&D) personnel and information and communications technology (ICT) readiness, then linked it to SDG scores. The strongest and most stable positive associations appeared for quality education, affordable and clean energy, and industry, innovation and infrastructure. Sustainable cities showed a positive but model-dependent link, while health showed a moderate full-model effect. Responsible consumption and climate action were not robust. Mechanism analysis pointed to human capital, industrial upgrading and total factor productivity (TFP) as positive channels, alongside lower energy intensity. Yet green innovation and government efficiency showed short-term friction, suggesting that early AI deployment can crowd out ecological innovation even as it improves operational efficiency. Low-Human Development Index (HDI) countries recorded larger marginal gains, especially in education, clean energy and sustainable production, but the authors caution that these gains depend on infrastructure, skills and governance. Scenario projections to 2100 found that a sustainability-oriented socioeconomic pathway combined with high AI development produced the most favorable SDG outcomes, yet regional divides persisted. Alternative specifications supported the main pattern.
The authors said the findings should temper claims that AI alone can deliver a net-zero transition. They said AI performs best where operations are data-rich and codifiable—energy access, industrial systems, education and urban management—but does not by itself change consumption patterns, regulatory ambition or decarbonization pace. They said the short-term drag on green innovation shows that deployment priorities and complementary institutions matter, and that low-HDI gains do not guarantee convergence without infrastructure, skills and clean electricity. They said AI is a conditional accelerator whose value depends on governance.
Policymakers should pair AI investment with structural climate policy, clean electricity, environmental regulation and institutional capacity, the authors argue. For developing and low-HDI countries, priorities include affordable digital infrastructure, AI-related skills, reliable power and access to clean energy technologies. For developed economies, green AI standards, low-carbon computing, responsible data governance and smaller digital footprints are central. International cooperation on technology transfer, open science and capacity building could help convert efficiency gains into more equitable transitions. The study also calls for sectoral and subnational analyses, distributional safeguards and dynamic assessments of efficiency–equity trade-offs. AI can support sustainable development when technological optimization is embedded in coordinated sustainability governance.
###
References
DOI
10.1016/j.ese.2026.100784
Original Source URL
https://doi.org/10.1016/j.ese.2026.100784
Funding information
National Natural Science Foundation of China (72504160); National Social Science Fund of China (25JYA04773); China Postdoctoral Science Foundation (2024M761575).
About Environmental Science and Ecotechnology
Environmental Science and Ecotechnology (ISSN 2666-4984) is an international, peer-reviewed, and open-access journal published by Elsevier. The journal publishes significant views and research across the full spectrum of ecology and environmental sciences, such as climate change, sustainability, biodiversity conservation, environment & health, green catalysis/processing for pollution control, and AI-driven environmental engineering. The latest impact factor of ESE is 14.3, according to the Journal Citation Reports TM 2024.