Send message to Authors

Paper short abstract

AI’s material infrastructures - data centers, chips, energy, and water systems - produce emissions and extract resources in ways affecting planetary and human health. Our scoping review maps scholarship connecting AI's environmental impacts to health outcomes, revealing key research gaps.

Paper long abstract

Artificial intelligence (AI) is widely promoted, yet the material infrastructures underpinning AI have significant consequences for planetary and human health. Data centers, high-performance computing (HPC) facilities, semiconductor manufacturing, energy grids, water-cooling, and other, less visible AI components are associated with environmental impacts including greenhouse gas emissions, water withdrawals, toxic releases, resource extraction, and heat discharge. These environmental impacts, in turn, are increasingly associated with health risks for humans on both individual and population levels. Structured overviews of these connections, however, are lacking.

This contribution maps the AI's environmental footprints' impacts on individual and public health. We explore how environmental risks of AI documented in peer-reviewed scholarship have been associated with health outcomes, both statistically and qualitatively. Drawing on findings from a scoping review of peer-reviewed studies published over the past 10 years, we present preliminary findings, including an overview of identified environmental impact categories, patterns of health linkage, and emerging methodological gaps.

These initial results open up discussion about how AI’s toxic world is enacted within peer-reviewed literature. The findings will shed light on how AI is already fostering new pathways for unequally distributing pollution and contamination, which, in turn, is creating unequal health risks for individuals and on a population level. As digitalization projects transform landscapes, these effects are at risk of intensifying. We will close by proposing avenues for participatory research to address these inequalities.