Environ Sci Ecotechnol. 2026 Oct 29;34:100784. doi: 10.1016/j.ese.2026.100784. eCollection 2026 Nov.
ABSTRACT
Global development agendas increasingly rely on artificial intelligence to accelerate sustainability transitions. As automated systems integrate into power grids and industrial operations, policymakers assume these technological upgrades will naturally translate into broader ecological gains. Yet it remains unclear whether algorithmic efficiency drives the deep structural changes needed for global climate action, or merely optimizes the status quo. Here we show an inherent efficiency-transformation gap in how artificial intelligence affects global sustainability. Analyzing 69 countries from 2000 to 2023, we find that technological diffusion robustly improves operational capacities-such as clean energy access (SDG 7) and industrial infrastructure (SDG 9)-driven by measurable gains in human capital and reduced national energy intensity. However, this techno-efficiency does not extend to systemic environmental governance. Artificial intelligence yields no robust progress in responsible consumption (SDG 12) or climate action (SDG 13). Instead, the rapid deployment of digital infrastructure temporarily crowds out green innovation, exposing a short-term friction between operational optimization and ecological transition. Furthermore, scenario projections to 2100 indicate that even under highly optimistic sustainability pathways, technological growth will not automatically erase regional development divides. Realizing the full promise of artificial intelligence requires policies that explicitly align computational power with structural climate action, rather than relying on efficiency gains to solve ecological crises.
PMID:43034684 | PMC:PMC13769063 | DOI:10.1016/j.ese.2026.100784