Those of us who work in the US healthcare system need a lot of things. A takeover by artificial intelligence companies isn’t one of them. Neither is the overwhelming hype that AI always seems to bring with it. Unfortunately, we’re getting both.

In January, for instance, The New York Times ran an op-ed with the headline “Stop Worrying, and Let AI Save Your Life.” The author, Dr. Bob Wachter, is the chair of the department of medicine at the University of California–San Francisco medical school and a proponent of AI’s use in healthcare. In the piece, Wachter touted AI as a way to fix the expensive inefficiencies of the American medical system, which he described as “in desperate need of transformation.” He will not be the last person to make such arguments.

As doctors who care for many people from marginalized communities, we find it difficult to be so bullish on AI—and not just because the quality and safety of AI technology is nowhere near guaranteed. We are worried that magical thinking about AI as a path toward real ‘transformation’ in healthcare, as Dr. Wachter suggests, will distract us from the current economic crisis that is crushing our patients. Far from its being a new chapter in how we deliver care, we fear that healthcare will become even more financialized as our systems become more deeply entangled with the AI bubble and the speculative investments that accompany it.

Medical journals now regularly feature robust debates about AI, and The New England Journal of Medicine and the Journal of the American Medical Association, the American medical journals with the largest footprint in medical science, now have entire sub-journals devoted to the topic. The coverage almost always deals with a narrow range of questions about the efficacy—and occasionally the medical ethics—of applying AI. How should we implement AI in [insert medical subspecialty]? Does AI perform as well as doctors on [insert task]? Occasionally someone will ask, “Does this really work as well as we think it does?” Questions about who stands to benefit most from the increased influence of tech profiteers over our healthcare system rarely get any space.

In the Times piece, Wachter hailed the use of AI assistants in clinical documentation. These so-called “scribes” are one of the simplest and most common forms of AI being systematically integrated in clinics across the country, including our own. The changes are currently marginal, but the conversation within medical institutions about them signals a trajectory toward stronger financial ties to the tech industry, with patients and healthcare workers as a captive “consumer” base and benefit accruing only to those who already hold political and economic power in our health system.

As doctors for patients who are poor, we think the current embrace of AI in healthcare misses the main issues at stake. Our patients are in the crosshairs of every form of violence produced by our current economic crisis. Many of them come from countries being decimated by war, sanctions, climate disasters, or all three. Here, they are hunted at home and at work by ICE agent; their hospitals are being closed; their jobs are disappearing; they are being evicted from their homes, and they are being cut off from SNAP and Medicaid benefits.

Current Issue

July/August 2026 Issue

A health system meant for the poor majority in this country must ask about how AI fits into that landscape of permanent crisis. The political economy of AI and the growth-focused ideology that governs its uptake across industries are poised to capture yet more control over our work. This capture will be more thorough if we believe that a rapid scale-up of AI will “save” the lives of anyone facing the circumstances that our patients currently face.

AI-related investments have accounted for a staggering 40 percent of US GDP growth in the past year. But much of this value is speculative, fueled by hundreds of billions in private capital investment and based on expected future gains rather than any existing substantial returns. Hoping to realize future profits, large tech firms employ a “scale at all costs” approach to building new data centers. This means buying land anywhere it can be found to build data centers that are hungry for energy and water, and employing precarious workers to comb through and label the most depraved content on the Internet.

Knowing that this AI boom is a house of cards built on a foundation of environmental decimation and exploitative labor practices, patients and patient-facing workers need to ask different questions about technology in our workplaces, and we should not surrender our power to shape how AI appears in the clinic.

For patients, healthcare providers, and health systems to understand the true implications of integrating AI into healthcare, we must be clear-eyed about the history and motivations for the introduction of technological interventions in the workplace more broadly. This extends beyond questions of the predictive power of any given algorithm, bias resulting from AI training on biased data sets, and even the question of how the FDA regulates (or in some cases doesn’t regulate) any given AI incursion into the clinic. As it stands, the goal of AI firms, especially the so-called Magnificent Seven who are the main sources of the growth in AI spending in recent years, is to make profit, not to cure illness or better the lives of regular people—“end users,” as they call us. They secure this profit by creating a technology that society (“the market” to capitalists) wants, needs, or accepts enough to incorporate into everyday life. Healthcare systems are moving forward with the adoption of AI not because of its potential to substantially improve care—and even that is an open question—but because these AI firms have created a market for something that only they can provide. In the market-based US healthcare system, institutions feel compelled to buy into any technology that might give them a competitive edge. But the only winners here are the firms developing and implementing this technology in the first place, and the healthcare executives whose profits are increasingly tied up with them.

Frighteningly, the dominant thought in healthcare seems to be that we need AI much more than it needs us. Health systems race to adopt new AI tools in the form of documentation assistants and clinical decision-making tools. More alarming for us are calls for AI firms to become state partners in Medicaid administration, a push dressed in the language of “health equity.” This kind of project, pushed by those with a flimsy track record of directly serving people on Medicaid, would give AI firms even more leverage in how providers deliver, and patients receive, care.

For clinicians like Wachter, the primary selling point seems to be around productivity and access. But increases in productivity almost always lead to an upward redistribution of wealth and record-breaking inequality, not improved wages or quality of life for the average person. Technology critics would likely caution us against the degradation of ambient AI tools once a critical mass of health systems have already bought in. Smaller tech companies that are currently responsive to patients, clinicians, and administrators may no longer have an incentive to respond to our needs if they are fully integrated and then acquired by a larger firm looking to cut corners and extract a profit. This outcome is also inevitable if AI exuberance remains unchecked in our current deregulatory environments.

In some ways, this fight is not new. Harold Waitzkin has described how the medical profession, through healthcare consolidations, de-skilling, and a general loss of autonomy over our work, has already been left “proletarianized” alongside our colleagues in other health professions and our patients. If we choose to become a passive social base for this technology, admitting large tech companies into our exam rooms without asking too many questions, we may soon find ourselves grappling with yet another layer of disempowerment. While some call AI “labor-saving technology,” we prefer the term “labor-replacing technology.” The latter more accurately describes the broader political economy of the past half-century.

Knowing this context, clinicians and patients should be asking a lot more questions about the integration of apparently innocuous AI tools. We don’t have to open the doors of our clinics based simply on a vague impression of inevitability, or at the behest of thought leaders who stand to personally benefit from their adoption. The “common sense” of AI in healthcare has been artificially inflated and propelled by investor enthusiasm, and the heavy-handed cheerleading of medicine’s most prominent thought leaders.