Little has captured public awareness in recent years more than the meteoric rise of artificial intelligence (AI). With this rise has come a fair share of controversy, with many both inside and outside of health care claiming, or at least hoping, that it will bring about a revolution that improves patient care, improves patient outcomes, reduces physician workloads, and drives down health care costs. Many are eager adopters, hopeful to make meaningful change, while others join hesitantly, having been warned that if they don’t adapt, they won’t survive. Even the American College of Preventive Medicine has started offering a certificate course on AI in public health and preventive medicine. Yet despite this growing adoption by physicians, scientists, and public health workers, who are typically conservative, cautious, and stringent when it comes to applying evidence-based methods and avoiding unknown or unnecessary risks to their patients, little of this same attention has been paid to the integration of AI. This is in spite of a growing body of literature warning of the ethical implications of AI use.
Although few would contest the potential of AI to be applied for either the benefit or the detriment of society, even if one were to make the herculean assumption that it was being used only for its benefit, there are still serious public health consequences inherent to all AI use that bear serious consideration and should give physicians and public health workers pause. Despite a marked lack of transparency concerning water use in the data centers powering AI, we still have ample evidence to suggest it is substantial and, perhaps even more importantly, that it is posing a direct and serious threat to the drinking water available to populations living in rural and desert communities. Furthermore, not only is water availability being threatened, but water quality is also at risk because of groundwater disruption and direct pollution of water sources.
The public health concerns don’t stop with water; air pollution from AI data centers poses a direct and serious threat to the communities they poison. Massive emissions of greenhouse gases are contributing to dangerous drops in local air quality near AI data centers, and the resulting health-related damages are anticipated to lead to premature deaths and tens of millions of dollars in health damages to nearby communities. Even the noise pollution from AI data centers is likely to lead to disability, cognitive decline, and increased mortality due to effects on cardiovascular health. The indirect costs of proximity to data centers, such as electricity price increases and grid failures and the risks associated with loss of cooling and refrigeration in summer, are also of serious public health concern.
Yet perhaps what should be most concerning to physicians and public health workers alike is whom these data centers affect most: vulnerable, at-risk, and minority populations. These data centers aren’t being built in wealthy gated communities; rather, they are being built next to underserved communities of color and rural populations with limited voices and avenues to fight back. The very communities that health care workers are working so hard to help, striving to overcome decades of institutional, structural, and political inequity, are the ones being harmed most by their own AI use. AI itself is already biased against them.
AI use is inherently damaging to the environment and, consequently, the health of any nearby population. In many cases, those nearby populations are already among the most vulnerable and in need of the most help. For those working in public health, using AI to further your goals is therefore akin to poisoning the well in search of a cure. The mantra of public health is prevention; it thus follows that every effort should be made to prevent the harms of AI use from occurring in the first place, rather than spending the next several decades studying the harms caused by data centers after the fact.
Despite the harms, it is unrealistic to assume that AI will ever disappear and unreasonable not to acknowledge that AI still has valuable use cases for public health and medicine at large. Therefore, only one viable path forward remains: regulation, policy reform, and strict and equitable environmental controls that can balance the harms of AI use with the potential benefits it can bring. As physicians and public health professionals, we have the ultimate responsibility of setting the highest possible standard of ethical and appropriate AI use, fighting back against all inappropriate uses, and even abstaining from AI use entirely until protections are put in place. Until then, our obligation is clear: “Do no harm.” We should therefore be at the forefront of advocacy efforts to protect the health of our patients and populations. An ounce of prevention is worth a pound of cure.
Jacob Player is a family medicine resident.
