In a perspective published in the Journal of Medical Internet Research, Jian-min Liu, MD, PhD, argues that despite rapid advances in artificial intelligence, the core of medical practice remains inherently human. The article identifies three domains where AI cannot replace physicians: the holistic sensory art of clinical observation and intuition; the longitudinal, trust-based doctor-patient relationship; and the capacity to accept clinical uncertainty and exercise moral responsibility. The author recommends preserving "AI-free clinical time" in medical training and safeguarding humanistic dimensions of care. The piece is framed for clinical doctors, medical students, educators, and health policy makers navigating AI integration in practice.
What happened
In a perspective in the Journal of Medical Internet Research, Jian-min Liu, MD, PhD, argues that although artificial intelligence and large language models have made substantial progress in areas such as image recognition and data mining, the practice of medicine retains irreducibly human elements. The article explicitly identifies three domains in which AI cannot replace doctors: the holistic sensory art of clinical observation and intuition; the longitudinal, trust-based doctor-patient relationship built on genuine emotional connection; and the capacity to embrace clinical uncertainty, exercise moral responsibility, and make courageous decisions. The perspective concludes by recommending preservation of "AI-free clinical time" in medical training and calls for safeguarding humanistic dimensions of care. The intended audience is clinical doctors, medical students, medical educators, and health policy makers.
Technical context
The article situates its argument against the backdrop of rapid technical improvements in multimodal AI and LLMs. Industry reporting and recent literature document gains in medical image analysis, nonimage data mining, and human-AI collaborative workflows; the perspective argues that technical competence alone does not capture the full remit of clinical care. A related June 2026 JMIR feature on LLM clinical reasoning notes that while o1-class models matched physician performance at ER triage on structured text, researchers found LLMs cannot replace a physician's ability to physically examine a patient, hear hesitation in their voice, or integrate messy information from multiple uncurated sources.
Broader context
The perspective aligns with a stream of 2026 clinical scholarship emphasizing patient-centered care and professional responsibility. A contemporaneous Family Medicine perspective by Stanford's Steven Lin, MD, frames five irreplaceable physician roles, from real-world guide to hands-on healer, reinforcing that AI is best positioned as a collaborative tool rather than a replacement. Medical educators and policy makers face competing pressures, efficiency gains from algorithmic decision support versus preserving human contact and moral accountability.
What to watch
Relevant signals include adoption of explicit curricular time labeled as "AI-free clinical time," empirical studies comparing patient outcomes and trust in AI-augmented versus human-led encounters, and regulatory guidance clarifying clinician responsibility when using AI-assisted tools.
Scoring Rationale #
An opinion perspective in JMIR arguing for the irreplaceable human elements of clinical practice amid AI advances. Directly relevant to AI-in-medicine practitioners and educators, but contains no new empirical data; contemporaneous peer perspectives from Stanford and JMIR itself reinforce the discourse. Solid but niche editorial piece rather than a major finding.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Health & Insurance problems