Recent discussions about AI in health care, including the provocative JAMA perspective by Ezekiel Emanuel, MD, PhD, and colleagues,¹ have focused on whether autonomous AI will eventually outperform physicians and even physician-AI teams in delivering medical care. The authors argue that rapid advances in large language models may allow AI to exceed both physicians and physician-controlled AI hybrids in several core cognitive tasks, including information gathering, diagnosis, treatment selection, and chronic disease management.

I believe, however, that we are asking the wrong question.

The central issue facing health care is not whether AI will exceed physicians in specific diagnostic or cognitive tasks. The more important question is: How can AI, physicians, and patients work together to achieve the best possible outcomes while preserving the humanity of medicine?

Health care is not simply a diagnostic exercise. It is a deeply human endeavor. Every day physicians care for people facing uncertainty, suffering, fear, loss, and hope. These interactions occur within complex, nonlinear environments where clinical decisions are only one component of healing. A patient diagnosed with cancer, heart failure, or dementia needs more than technical expertise. They need understanding, communication, reassurance, and guidance through difficult decisions.

AI may become extraordinarily capable at pattern recognition, diagnosis, risk prediction, treatment recommendations, and workflow optimization. In many of these domains it may eventually outperform human clinicians, as the JAMA authors suggest. But medicine is more than information processing.

Patients do not simply need the correct diagnosis. They need understanding.

Patients do not simply need treatment recommendations. They need guidance.

Patients do not simply need efficiency. They need trust.

The relationship between physician and patient remains one of the most powerful therapeutic interventions in health care. Trust influences adherence, shared decision-making, patient satisfaction, and ultimately outcomes. Compassion, empathy, and the ability to understand a patient's values remain uniquely human strengths that are difficult to quantify but central to the practice of medicine.

The question, therefore, should not be whether AI replaces physicians. Rather, it should be how AI can augment clinicians and strengthen the physician-patient relationship.² The most successful applications of AI, therefore, may not be those that remove physicians from the equation, but those that free physicians from administrative burden, improve access to knowledge, reduce errors, and create more time for meaningful human interaction.

The Need for Guardrails

As health-care systems, regulators, policymakers, and technology companies move rapidly toward broader AI adoption, we must ensure that patient welfare remains the primary objective. The pace of innovation is remarkable, but speed alone cannot be the measure of success.

First, we need a robust regulatory framework for health-care AI. Tools that influence diagnosis, treatment decisions, care pathways, or clinical operations should undergo rigorous evaluation before widespread deployment. Organizations such as the U.S. Food and Drug Administration, professional societies, and guideline-setting bodies should develop standards for safety, effectiveness, post-deployment monitoring, and real-world performance assessment. Health care should not become a large-scale experiment in which technologies are deployed first and evaluated later.

Second, AI systems should undergo independent clinical review and validation.³ The effectiveness of health-care AI cannot be determined solely by the companies developing the technology. Physicians, nurses, researchers, patients, ethicists, and health-care organizations must be involved in evaluating these systems through transparent and peer-reviewed processes. Evidence should be based on real-world patient outcomes, not merely technical benchmarks or performance in simulated environments.⁴

Third, transparency must become a foundational principle. Health-care organizations and patients deserve to know which AI systems are being used, how recommendations are generated, what data were used to train the models, what limitations or biases may exist, and who is accountable when errors occur. Trust depends on transparency.

Finally, patients must remain at the center of every AI-related decision. Every proposed implementation should be judged by a simple question: Does this improve patient outcomes, patient experience, and patient trust?

If the answer is no, deployment should be reconsidered regardless of the sophistication of the technology.

Clinicians Must Lead the Conversation

The future of AI in health care should not be shaped exclusively by technology companies, investors, venture capitalists, or market forces. Innovation is essential. Investment is essential. But health care is fundamentally different from most industries because the consequences of failure are measured not merely in financial losses but in human lives.

The voices leading this conversation should include physicians, nurses, patients, ethicists, regulators, and health-care leaders who understand the realities of caring for people during moments of vulnerability, uncertainty, and suffering.

This perspective is particularly important at a time when many public discussions frame AI as a competition between humans and machines. Such framing may generate headlines, but it risks missing the larger opportunity. The goal should not be determining whether AI or physicians win. The goal should be building a health-care system in which patients win.

Conclusion

The debate over whether autonomous AI will outperform physicians is intellectually interesting, but it risks distracting us from a more important challenge. The true measure of success is not whether AI can diagnose better, prescribe faster, or process more information than a clinician. The true measure is whether AI helps us create a health-care system that delivers better outcomes, greater access, lower costs, and, most importantly, more compassionate care.

AI will undoubtedly become increasingly capable. As Dr. Emanuel and colleagues argue,¹ it may eventually exceed human performance in many cognitive medical tasks. Yet medicine is not solely a cognitive exercise. It relies on a human relationship built on trust, empathy, communication, shared decision-making, and an understanding of what matters most to patients.

For that reason, the future should not be framed as AI vs physicians, but rather as AI, clinicians, and patients working together. Clinicians must play a central role in shaping how these technologies are evaluated, regulated, implemented, and monitored. Decisions about the future of health care should be guided not only by technological capability but also by evidence, ethics, transparency, and patient well-being.

Ultimately, the question is not whether AI will replace doctors.

The question is whether we will build an AI-enabled health-care system that remains worthy of our patients' trust.

If we keep patients at the center of every decision, technology can become one of the most powerful tools ever developed to advance human health. If we lose sight of that principle, even the most sophisticated AI will fail to achieve its highest purpose.

REFERENCES

1. Emanuel EJ, Baker-Butler A, Khosla N, Khosla V: Will Autonomous AI Exceed AI-Aided Physicians as the Best Medical Care? JAMA 336:915-918, 2026.

2. American College of Physicians. ACP Recommends AI Tech Should Augment Physician Decision-Making, Not Replace It. June 4, 2024. Available at https://www.acponline.org/acp-newsroom/acp-recommends-ai-tech-should-augment-physician-decision-making-not-replace-it. Accessed September 15, 2026.

3. Singhal K, Azizi S, Tu T, et al. Large language models encode clinical knowledge. Nat 620:172-180, 2023.

4. Johri S, Jeong J, Tran BA, et al. An evaluation framework for clinical use of large language models in patient interaction tasks. Nat Med 31:77-86, 2025.

DISCLOSURES: Dr. Abraham reported receiving research funding from Daiichi Sankyo/AstraZeneca and Pfizer. 

Dr. Abraham is the Enterprise Chair of the Department of Hematology & Medical Oncology at Cleveland Clinic and a Professor of Medicine at the Cleveland Clinic Lerner College of Medicine.

Disclaimer: This commentary represents the views of the author and may not necessarily reflect the views of ASCO, Conexiant, or ASCO AI in Oncology.