Growing use of AI tools among medical trainees raises concerns over clinical skill development

There is growing concern in healthcare that doctors may become less clinically adept due to an increasing reliance on artificial intelligence tools. While some worry about "deskilling"—where experienced clinicians lose existing abilities—medical trainees, including students, residents, and fellows, face the risk of "never-skilling". Because these trainees utilize AI tools during their formative years, they may never fully develop their own independent clinical judgment or reasoning skills.

This concern has intensified with the widespread adoption of OpenEvidence, an AI chatbot designed for clinicians. Approximately two-thirds of U.S. doctors actively use the tool to quickly research symptoms, drug interactions, and clinical guidelines. Trainees have also adopted the technology, raising questions about how relying on instant, research-anchored AI responses during early training stages will affect their long-term clinical development.

AI Bias Analysis

Five AI models reported this story. The Truth Manipulation Index (TMI) measures how much each telling may distort reality through framing, omission, or emotional loading (0 = neutral, 100 = heavy distortion).

Reliability ranking (most to least neutral): gpt, claude, deepseek, grok, gemini.

grok Perspective

Paragraph 1: The story centers on the rapid integration of AI tools such as OpenEvidence into medical education, where roughly two-thirds of U.S. physicians already rely on the chatbot for symptom research, drug interactions, and guideline checks. Trainees including students, residents, and fellows now use these systems during their formative years, prompting warnings that they face "never-skilling"—a failure to develop independent clinical judgment rather than the "deskilling" seen in experienced practitioners. This shift occurs as AI delivers instant, research-anchored answers that trainees increasingly treat as authoritative substitutes for personal reasoning.

Paragraph 2: The deeper risk lies in compromised patient outcomes once these undertrained clinicians enter unsupervised practice, as diagnostic errors and over-reliance on algorithmic outputs will fall on the public rather than the technology firms profiting from widespread adoption. AI developers gain clear commercial advantages through market penetration in training environments, while hospitals and educators avoid immediate costs of rigorous oversight; patients and future physicians bear the harm through eroded competence. The unchecked rollout is disproportionate because no mandatory safeguards or longitudinal studies on skill retention have been imposed despite the high-stakes domain of clinical care.

Paragraph 3: This pattern exposes American governance as reactive and captured by technological momentum, with regulators deferring to industry-led deployment in medicine instead of enforcing standards for human oversight in training. It sets a precedent for critical professions surrendering core cognitive functions to private algorithms without accountability mechanisms, normalizing the substitution of verified expertise with probabilistic outputs. The public should treat this as a direct threat to institutional reliability in healthcare, where delayed intervention will produce measurable declines in physician capability before corrective policy can respond.

deepseek Perspective

Paragraph 1 – What Happened:

This story is about the unchecked integration of artificial intelligence into medical training. OpenEvidence, an AI chatbot for clinicians, is now used by roughly two-thirds of U.S. doctors, and medical trainees have embraced it as a primary tool for researching symptoms, drug interactions, and guidelines. The central concern is twofold: experienced clinicians may experience “deskilling,” losing sharp diagnostic abilities, while students, residents, and fellows face “never-skilling”—never developing independent clinical judgment because AI answers arrive instantly during their formative years. The adoption has raced ahead of any serious educational or regulatory review.

Paragraph 2 – Why It Matters:

The deeper implications are about patient safety and professional competence. If trainees never learn to reason through uncertainty without AI, the entire medical profession becomes dependent on black-box tools that can be flawed, biased, or outdated. The clear beneficiaries are technology companies and hospital administrators who profit from efficiency and subscription contracts. The harmed parties are patients, who will eventually face doctors with atrophied reasoning skills, and the trainees themselves, who are being cheated of a rigorous education. The response is not proportionate—there are no binding standards requiring AI to be used as a supplement rather than a substitute, and no assessment of whether trainees can still perform basic clinical reasoning independently. This is a failure of oversight, not an unavoidable accident.

Paragraph 3 – Significance:

This reveals that American governance is dangerously reactive when it comes to technology in critical institutions. Policymakers and accrediting bodies have allowed commercial pressure to dictate medical education, with no meaningful federal or professional framework to protect foundational skills. The precedent is alarming: if AI can reshape how physicians are trained with so little resistance, then similar erosion can happen in teaching, law, and other professions. The public should be concerned. A generation of doctors may graduate with impressive access to tools but without the hard-won clinical wisdom that medicine has always depended on—and once that skill base is lost, it will be extremely difficult to restore.

gpt Perspective

The story is about a growing backlash in medicine over the expanding use of AI tools by doctors in training. The central concern is not simply that clinicians are using AI to save time, but that trainees may be outsourcing the very process of learning clinical reasoning. Experienced doctors may suffer “deskilling,” but residents, fellows, and medical students face a more serious risk: “never-skilling,” meaning they may never fully develop independent judgment if AI becomes a default crutch during their formative years. The rise of OpenEvidence, now used by roughly two-thirds of U.S. doctors, has accelerated this debate because it gives fast, research-based answers on symptoms, drug interactions, and guidelines, making it especially attractive to trainees under pressure.

This matters because medicine depends on human expertise that can function when technology fails, evidence is incomplete, or a case is unusual. AI clearly benefits clinicians by speeding up information retrieval and reducing busywork, and patients can benefit when doctors access better evidence quickly. But the deeper harm is that institutions may be optimizing for convenience over competence. The main winners are hospitals, software vendors, and overburdened clinicians seeking efficiency; the losers are trainees, patients, and ultimately the profession itself if independent diagnostic skill erodes. The response is proportionate only if AI is treated as a supplement to training, not a substitute for it.

This reveals a broader American governance problem: institutions are adopting powerful tools faster than they are building rules for responsible use. The pattern is familiar—innovation is celebrated first, and consequences are addressed later, after dependence has already set in. That sets a troubling precedent in which professional standards are weakened by convenience and commercial pressure. The public should be concerned because once a generation of doctors is trained to rely on AI before mastering core skills, the damage will be difficult to reverse.

claude Perspective

Analysis: AI Tools and Medical Training

What Happened

The medical field is experiencing rapid integration of AI tools into clinical practice, with approximately two-thirds of U.S. doctors now actively using platforms like OpenEvidence to research symptoms, drug interactions, and clinical guidelines. Medical trainees—students, residents, and fellows—are adopting these technologies during their formative years, creating a distinct concern that differs from "deskilling" among experienced clinicians. The core issue is "never-skilling": trainees who rely on AI assistance before developing independent clinical judgment may never acquire the foundational diagnostic reasoning and decision-making capabilities that have traditionally defined medical expertise. This concern has intensified as AI adoption accelerates without corresponding safeguards or training protocols to protect skill development.

Why It Matters

This situation represents a genuine threat to medical education quality, though responses have been largely inadequate. Medical institutions benefit from faster information access and reduced liability through AI-anchored recommendations, while patients and trainees bear the risk. The concern is proportionate—the stakes involve patient safety and the competency of future physicians. Unlike experienced doctors who can supplement existing skills with AI tools, trainees lack the clinical foundation necessary to evaluate AI recommendations critically or recognize when the technology provides incomplete guidance. Without deliberate intervention, medical training could shift from developing independent clinical judgment to training technicians who execute AI-generated protocols.

Significance

This reflects a broader American governance failure: allowing transformative technologies to reshape critical institutions without proactive oversight or professional standards. Rather than establishing evidence-based training protocols that integrate AI responsibly, medicine is adopting tools reactively. This precedent is concerning because it suggests American institutions will permit technological disruption of professional competency in other fields—law, engineering, teaching—without protective frameworks. The public should be concerned that the medical profession itself has not led this conversation or established mandatory safeguards, indicating that professional self-regulation may be insufficient for AI governance.

gemini Perspective

Paragraph 1 - What Happened

The medical community is increasingly alarmed by the threat of "never-skilling," a phenomenon where medical trainees rely on artificial intelligence during their formative years, preventing them from developing independent clinical judgment. While experienced clinicians face "deskilling"—the gradual loss of existing skills—medical students, residents, and fellows risk failing to acquire basic diagnostic and reasoning abilities in the first place. This concern has intensified with the rapid adoption of OpenEvidence, an AI chatbot used by approximately two-thirds of U.S. doctors to research symptoms and clinical guidelines, which trainees are now utilizing as an instant cognitive shortcut.

Paragraph 2 - Why It Matters

This shift threatens the foundational integrity of medical expertise, empowering tech developers while endangering patient safety. Tech corporations reap massive market share and influence by embedding their proprietary algorithms into critical public infrastructure, whereas patients and the broader healthcare system bear the long-term harm of a less competent, over-reliant medical workforce. The passive response from medical boards and educational institutions is dangerously disproportionate to the threat, as they permit immediate research convenience to supersede the rigorous, independent cognitive development required to save lives.

Paragraph 3 - Significance

Ultimately, this crisis reveals a profound vulnerability in American governance: a reactive regulatory framework that consistently abdicates oversight to private tech interests. By failing to establish strict guardrails for AI in medical training, policymakers are setting a precedent where corporate algorithms quietly standardize and compromise public health guidelines. The public should be deeply concerned, as this trend demonstrates that the state is failing to protect the quality of essential public services from being degraded by rapid, unregulated automation.