The impacts of artificial intelligence on environmental sustainability and human well-being
A massive review finds 83% of environmental studies see AI as positive, while well-being impacts are split.
A new systematic literature review, 'The impacts of artificial intelligence on environmental sustainability and human well-being,' authored by Noemi Luna Carmeno, Tiago Domingos, and Daniel W. O'Neill, provides a comprehensive analysis of AI's dual-edged effects. The study, which synthesized findings from 1,291 selected papers out of 6,655 records, reveals a significant research imbalance. It found that 83% of environmental impact studies portray AI's effects as positive, but these studies are overwhelmingly narrow, with 72% focusing only on energy use and CO2 emissions while largely ignoring systemic effects on water, materials, and biodiversity. Conversely, research on human well-being is more conceptual and shows a near-even split between positive and negative outcomes, masking critical differences across social dimensions.
The technical breakdown shows a stark contrast in perceived impacts. While AI is expected to positively influence income and health, its effects on inequality, social cohesion, and employment are predominantly viewed as negative. The authors argue that current assessments are fragmented, failing to integrate computing-related, application-level, and systemic impacts across both environmental and social spheres. To steer AI development toward genuine sustainability and human flourishing, the study urgently calls for future research to adopt full life-cycle environmental perspectives and prioritize empirical, multidimensional well-being analyses. Bridging these critical gaps is essential for policymakers and developers to understand and mitigate AI's complex, real-world consequences.
- Environmental research is skewed: 83% of studies show positive impacts, but 72% focus narrowly only on energy and CO2, missing systemic effects.
- Well-being impacts are split: Overall, 44% of studies are positive and 46% negative, with AI expected to boost income/health but worsen inequality and jobs.
- The review of 1,291 studies calls for integrated research covering full life-cycle environmental costs and empirical social analysis to guide AI development.
Why It Matters
This meta-analysis exposes critical blind spots in how we measure AI's true costs and benefits, essential for responsible development.