TLDR: A research paper by Robert Sparrow and Joshua Hatherley challenges Eric Topol’s optimistic view that AI in healthcare will enable doctors to provide more empathetic care. They argue that economic pressures, institutional focus on efficiency, potential disempowerment of physicians, increased data burdens, and the “black box” nature of AI are more likely to erode doctor-patient relationships and trust. The authors conclude that addressing the underlying economic and institutional context of medicine is crucial to ensure AI truly benefits patient care, rather than just increasing efficiency.
In his widely discussed book, “Deep Medicine,” Eric Topol presents an optimistic vision for the future of healthcare, suggesting that artificial intelligence (AI) will revolutionize medicine. He posits that AI will become sophisticated enough to handle many routine tasks, thereby freeing physicians to dedicate more time and attention to providing empathetic, human-centered care to their patients. Topol believes AI will offer comprehensive views of patient data, improve decision-making, reduce errors, and guide treatment, ultimately restoring the precious doctor-patient connection.
However, a research paper titled “High hopes for Deep Medicine? AI, economics, and the future of care” by Robert Sparrow and Joshua Hatherley offers a contrasting, more cautious perspective. They argue that several factors suggest a radically different future for healthcare, one where AI might further erode therapeutic relationships and diminish both professional and patient satisfaction.
One of the primary concerns highlighted by Sparrow and Hatherley is the significant role of economic and institutional considerations. In for-profit healthcare systems, any time saved by AI reducing administrative burdens is likely to be used to increase patient throughput rather than allowing doctors more time with individual patients. Even in the public sector, the drive for cost savings and efficiency will likely lead to AI being used to expand access by treating more people per day, rather than deepening the time spent with each patient. Institutions tend to prioritize measurable metrics like “time per patient” over the more subtle and intangible aspects of “care,” reinforcing this trend.
The authors also point out that the introduction of AI is likely to be highly disruptive to the medical profession. While AI may not entirely replace physicians, it could render certain skills redundant, shift power away from frontline practitioners to IT system managers, and even lead to increased surveillance of physician performance. A demoralized, disempowered, and surveilled medical profession is ill-equipped to advocate for better working conditions or more patient-centered care, as Topol hopes they might.
Historically, doctors have struggled to resist major changes like the rise of managed care or the problematic implementation of electronic medical records (EHRs), which often increased administrative burdens rather than alleviating them. This history provides little confidence that the medical profession will be able to resist similar economic and institutional pressures when it comes to AI adoption.
Furthermore, the paper suggests that AI might actually increase, rather than decrease, the administrative load on healthcare workers. AI thrives on big data, meaning the demand for measuring and recording information in medical settings will likely intensify. This could lead to healthcare workers spending more time staring at screens and filling out forms, diverting their attention from patients. While AI might automate some data management, the complex nature of real-world patient interactions, including nuances like accents, body language, and social context, still requires significant human judgment to interpret and integrate data.
Finally, a profound tension exists regarding trust. AI often functions as a “black box,” where the reasoning behind its outputs is not easily understood, even by its designers. If doctors rely on AI for diagnoses and treatment recommendations without fully understanding the rationale, it raises questions about how patients can trust their doctors. If patients perceive their physician as merely a “handmaiden” to an AI, the sense of being cared for by a human professional could diminish. While AI offers benefits like timely and accurate diagnoses, these advancements may come at a cost to the human element of care, especially given existing pressures on healthcare providers.
Also Read:
- Beyond the Buzz: Understanding Large Language Models in Medicine
- The Hidden Cost of Empathetic AI: A Trade-Off in Reliability
Sparrow and Hatherley conclude that simply wishing for AI to expand opportunities for care is insufficient. The factors that have led to the decline in human contact in medicine are primarily economic and political. Therefore, ensuring AI enhances rather than erodes care requires a deeper examination of the business of medicine and the institutional and economic contexts in which it operates today, rather than solely focusing on technological advancements.
