Artificial intelligence: what physicians say, in their own words

Last updated September 20, 2026.

This page is a maintained record of what physicians have written about artificial intelligence in medicine on KevinMD.com, a physician-authored publication founded in 2004 by Kevin Pho, MD, a board-certified internal medicine physician in Nashua, New Hampshire. It draws on 517 essays and podcast transcripts with AI, artificial intelligence, ChatGPT, machine learning, a large language model, generative AI, or an algorithm in the title, published between April 2006 and September 2026 by 276 named contributors, 298 of the posts bylined by physicians. Of the 517, 470 were published from 2023 onward and 207 in 2026 alone, which makes this the fastest-growing subject in the archive. Every claim is attributed to a named author with the month it was published, quoted in the author’s own words, and linked to the original.

AI scribes and documentation have their own record on the electronic health record page; insurers’ use of AI to deny care is on the prior authorization page. This page takes the rest. Each section opens with a direct answer, then what named authors have said, in the order they said it. Where authors cite studies or figures, the source is named alongside the author and the figures are theirs. Contributors who build or sell AI tools are identified as such. This is physician opinion and experience, not a product guide or a guideline.

What do physicians say AI can do in clinical practice now?

Physicians writing on KevinMD describe three things the tools do reliably as of September 2026: draft documentation, read images alongside a radiologist, and surface a diagnosis the physician might have missed. Everything beyond that they call promising, unvalidated, or both. Posts mentioning diagnosis have been the majority in every period, 57 percent in 2026.

Michael Kirsch, MD, wrote in September 2018 that a program fed a patient’s symptoms and labs “might alert me to diagnostic possibilities that are beyond my reach,” in “Will artificial intelligence ever replace doctors?” Jennifer Lycette, MD, an oncologist, wrote in April 2023 that the question that keeps oncologists up at night, “Could I have done something more, or differently,” is not one a precision medicine report answers, in “I’m a medical oncologist. Here’s why AI isn’t going to cure cancer.” Hannah Kopelman, DO, MBA, wrote in July 2023 that a language model “could be used as a triage service, where primary care physicians share their observations,” in “How artificial intelligence (AI) is transforming dermatology diagnosis and treatment.” Harvey Castro, MD, MBA, the most frequent contributor on the subject and an AI consultant, wrote in March 2025 that one vendor had reported “a reduction in documentation time of approximately 41 percent, facilitating increased patient interaction,” in “AI agents in health care: Balancing innovation, trust, and evidence-based implementation.” Ara Feinstein, MD, MPH, wrote in September 2025 that Microsoft’s diagnostic system had reportedly “diagnosed complex cases from the New England Journal of Medicine with eighty percent accuracy,” in “The parallel evolution of computer chess and AI in health care: the inevitable journey to embracing cognitive inferiority.” Sujay Jadhav, a health care executive, said on The Podcast by KevinMD in October 2025 that “we are mining the physician notes and the images to help do early diagnosis,” in “How AI is transforming health care with real-world data insights [PODCAST].” Neil Panchal, DO, wrote in June 2026 that “81 percent of physicians are already using some form of AI at work,” and that poorly calibrated alerts are “worse than no alert at all,”, in “AI tools for physicians: What to use, what to ignore.”

Questions on this page · All records

Will AI replace physicians? Physicians disagree.

Most physicians writing on KevinMD say no, and they have said it the same way since 2015: the tool pattern-matches, the physician makes sense of a person. Posts mentioning replacement held below 8 percent of posts until 2024, then rose to 21 percent in 2025 and 2026, and the 2026 answers are less about accuracy than about who is accountable and what gets paid for.

Nathan A. Pennell, MD, PhD, an oncologist, wrote in December 2015 that he had seen “patients with NSCLC who were actively dying come back from the brink,” on a targeted drug, in “This is why a computer algorithm cannot ever fully replace a doctor’s judgment.” Bruce Campbell, MD, wrote in July 2017 that a program might someday hold the conversations that follow a diagnosis, but “it will be a long, long time before it can,” in “Artificial intelligence in medicine is still a long ways off.” Richard Plotzker, MD, wrote in April 2018 that some patients “need a little more than just being processed through in the shortest length of stay,” in “Computer algorithms don’t think like doctors.” Arthur Lazarus, MD, MBA, wrote in December 2024 that even as “multiple studies have shown LLMs outscoring human physicians,” there is “no substitute for the human touch,” in “AI’s cognitive gap: Why human doctors remain irreplaceable.”

Questions on this page · All records

Is AI already replacing physicians? Physicians disagree.

A minority of KevinMD’s physician authors say the replacement question is being answered by economics rather than software: by systems that substituted protocol for experience years ago, by patients who search before they book, and by procedural revenue that quietly subsidizes the human work. One proposes semi-autonomous AI primary care outright.

Steven Charlap, MD, MBA, wrote in October 2024 that “the concept of semi-autonomous AI-powered primary care providers emerges as a promising solution,” in “Towards a future for semi-autonomous AI-powered primary care providers.” Bhargav Raman, MD, MBA, wrote in July 2025 that “the industry has already decided to replace doctors with those with less experience,” in “AI is already replacing doctors, just not how you think.” P. Dileep Kumar, MD, MBA, wrote in January 2026 that “this replacement will not be done by any fancy robots, but interestingly by the patients themselves,” in “Will AI replace primary care physicians?” Ronald L. Lindsay, MD, wrote in August 2026 that “The fear of replacement is really the fear of accountability,” in “Medicine disposed of physicians long before it feared AI.” Andrey Espinoza, MD, a cardiologist, wrote in September 2026 that if AI takes the procedural work, “the human work may simply stop being paid for,” in “The real risk of AI replacing doctors is not accuracy.”

Questions on this page · All records

Will AI replace radiologists?

Radiology is the specialty physicians on KevinMD name most often, and its radiologists have written the sharpest rebuttals: the authorized tools are assistive by regulatory definition, vendor studies flatter the software, and the visible cost so far is students steering away from the field. Posts mentioning radiology or imaging have held between 24 and 30 percent of posts since 2017.

Brett Mollard, MD, wrote in May 2023 that “Deep learning algorithms are already learning from various imaging repositories and pathology databases,” in “How AI is enhancing patient care and improving radiologists’ lives.” Fardad Behzadi, MD, a radiologist, wrote in June 2025 that of radiology AI studies, “The data is almost always selectively presented to support a pre-determined narrative,” in “AI is not a threat to radiologists. It’s a distraction from what truly matters in medicine.” Dr. Robert Wachter, chair of medicine at the University of California, San Francisco, said on The Podcast by KevinMD in April 2026 that Geoffrey Hinton in 2016 “infamously said we should stop training any new radiologists,” in “Gradually, then suddenly: Dr. Robert Wachter on health care’s giant AI leap [PODCAST].” Dr. Amarachi Amanda Dukor, a physician in Nigeria, wrote in May 2026 that AI “is often positioned as the solution to the field’s most pressing challenges,” in settings where the shortage is of scanners and radiologists, in “Artificial intelligence in radiology and global health equity.” Pouyan Golshani, MD, an interventional radiologist, wrote in June 2026 that of 1,430 FDA-authorized AI devices, “76.5 percent were reviewed by the FDA’s Radiology panel,” and that “A third-year medical student told me last month she was not going to apply to radiology”, in “Is AI replacing radiologists? A radiologist disagrees.”

Questions on this page · All records

Who is responsible when AI is wrong?

On responsibility, physicians on KevinMD describe one settled fact and one open question. Settled: the physician who signs is responsible, whatever the software contributed. Open: whether developers and institutions will come to share that responsibility, and whether following a tool against the standard of care is defensible. Posts mentioning responsibility or accountability rose from 28 percent of 2024 posts to 54 percent in 2026.

Arthur Lazarus, MD, MBA, wrote in January 2026 that the January 2026 FDA guidance “places liability back in the physician’s hands, but influence is harder to regulate than responsibility,” in “FDA loosens AI oversight: What clinicians need to know about the 2026 guidance.” Ian Hu and Pao Hsuan Huang wrote in April 2026 that disclaimers appear when a tool is wrong, but “in the real world, patients act on these outputs,” in “Why accountability in medicine must guide health care AI.” Erin J. Silvertooth, MD, wrote in June 2026 that every software license says “if the tool makes a mistake, the clinician who used it is responsible for what happens,” in “Clinical AI liability lands on you, not the vendor.” Richard E. Anderson, chief executive of a physician-owned malpractice insurer, said on The Podcast by KevinMD in June 2026 that if you follow a tool against the standard and the outcome is bad, “you have actually deviated from the standard of care to follow AI,” in “Why AI has outpaced medical malpractice law, and what to do about it [PODCAST].” Teddy A. Teddy, MD, a cardiologist, wrote in September 2026 that “current malpractice law generally assigns primary responsibility to the physician regardless of whether AI contributed,” in “When AI misses a diagnosis, the cardiologist signs alone.” Brian Hudes, MD, who is building a clinical AI platform, wrote in September 2026 that the tools’ omission rates are “knowable, testable, and currently not disclosed to the clinician who signs the note,” in “Why clinical AI limitations are not a physician problem.”

Questions on this page · All records

What do physicians say about AI errors and hallucinations?

KevinMD’s authors documented the failure modes in the order they found them: a recommendation that ignored a black box warning in 2018, invented facts since 2023, fabricated references in 2026. Posts mentioning error, hallucination, mistake, or wrong rose from 43 percent of 2024 posts to 63 percent in 2026.

Skeptical Scalpel, MD, wrote in August 2018 that Watson recommended, for a mock lung cancer patient with bleeding, “a drug which carries a black box warning saying its use can cause massive hemorrhage,” in “Artificial intelligence in medicine: not ready for prime time.” Harvey Castro, MD, MBA, wrote in July 2023 that “Hallucinations generally occur when the AI model attempts to generate a response, regardless of whether it has the necessary knowledge,” in “Lowering hallucination results with ChatGPT.” L. Joseph Parker, MD, wrote in January 2024 that the Justice Department had “committed twelve federal prosecutors to enforce the AI’s decisions,” while experts diagnosing chronic pain “only agree 43.0 percent of the time,”, in “AI enforcement in health care: Unpacking the DEA’s approach to the opioid epidemic.” Neil Anand, MD, wrote in December 2024 that accuracy is the most common metric and “relying solely on one AI performance metric can result in inefficiencies or even dangerous outcomes,” in “Navigating Goodhart’s Law dilemma and the future of AI in medicine.” P. Dileep Kumar, MD, MBA, wrote in April 2025 that a 2023 federal class action against UnitedHealthcare Group “alleges a 90 percent error rate in AI programs used for prior authorizations,” in “AI in health care: the black box of prior authorization.” Rao M. Uppu, PhD, wrote in July 2026 that in medicine “a fabricated or poorly verified reference is not simply a bibliographic error,” in “AI fabricated references threaten medical evidence.”

Questions on this page · All records

What do physicians say about patients who arrive with an AI answer?

A subject with no precedent on KevinMD before 2023 except Dr. Google. Its physicians describe patients anchored on a diagnosis a 15-minute visit cannot unwind, patients who trust a tool because it felt understanding, and, in 2026, psychiatric patients whose delusions a chatbot reinforced. Posts mentioning chatbots rose from 10 percent of 2024 posts to 22 percent in 2025 and 2026.

Robert Pearl, MD, wrote in December 2023 that “generative AI will alter the doctor-patient dynamic by leveling the playing field,” in “How generative AI will upend the doctor-patient relationship.” Muhamad Aly Rifai, MD, a psychiatrist, wrote in May 2026 that a therapy chatbot sounds authoritative while having “no license, no patient relationship, no duty, no real judgment, and no accountability,” in “AI therapy chatbots are crossing into impersonation.” Ganesh Asaithambi wrote in June 2026 that for AI-anchored diagnoses, “visit structures designed around 15-minute problem-focused encounters are poorly equipped to address them,” in “When the AI diagnosis arrives before the patient does.” Robert Pak, MBA and Thomas Pak, MD, PhD, wrote in August 2026 that in a 2025 study of 29 chatbots on suicidal ideation, “none of the chatbots met its criteria for an adequate response,” in “AI chatbots and suicide: what our tests found.” Simon Mathews, MD, wrote in August 2026 that Gallup polling had found that “one in seven recent AI health users skipped a doctor’s visit based on what a chatbot said,” in “Before you paste your labs into ChatGPT, read this.” Timothy Lesaca, MD, wrote in September 2026 that the drift medicine should notice runs “from helpfulness to perceived understanding to trust in what the system says,” in “Patient trust in AI is not proof the advice is right.” Nicole Drapeau Gillen wrote in September 2026 that a King’s College London team had found that “AI psychosis looks clinically different from schizophrenia,” in “AI psychosis is not schizophrenia. The difference matters.”

If you or someone you know is struggling, call or text 988 in the United States. Physicians can call the Physician Support Line at 1-888-409-0141.

Questions on this page · All records

Can AI be more empathetic than a physician? Physicians disagree.

KevinMD’s physician authors split on the empathy studies. Some accept that chatbot answers can read as warmer than a rushed clinician’s and propose using the tool to extend empathy; others call the warmth an illusion of care and say efficiency displacing empathy is the danger. Posts mentioning empathy or the human element have held near a third of posts since 2024.

Harvey Castro, MD, MBA, wrote in June 2025 that AI may diagnose faster, but “it will never replace the human connection that defines medicine,” in “AI and humanity in health care: Preserving what makes us human.” Vanessa D’Amario, PhD and Vijay Rajput, MD, wrote in September 2025 that in a model they call augmented empathy, “AI tools support, rather than replace empathetic care,” in “Is it time to embrace augmented empathy while using artificial intelligence in health care?” Ronke Lawal, MBA, wrote in October 2025 that the test of a therapeutic AI is “Does it build capacity for genuine human connection, or does it allow users to avoid the discomfort,” in “AI companions and loneliness.” Gerald Kuo wrote in March 2026 that “AI can simulate empathetic language, but it cannot experience authentic concern,” in “The hidden dangers of AI voice assistants in elder care.” Cybil Sierra Stingl, MD and Robert M. Kaplan, PhD, wrote in April 2026 that a 2023 JAMA Internal Medicine study found that “clinicians rated chatbot responses to real patient questions as higher quality and more empathetic,” in “How generative AI in health care is changing patient expectations.”

Questions on this page · All records

What do physicians say about bias and equity in medical AI?

Physicians on KevinMD have written about bias in medical AI since 2018, and the examples are concrete: a risk score that rated Black patients healthier than white patients at the same level of illness, a kidney formula with a race correction, dermatology models trained on light skin. In 2026 the argument widened to who supervises AI where there are no specialists. Posts mentioning bias or equity peaked at 46 percent in 2025.

Iyesatta Massaquoi Emeli, MD, an emergency physician, wrote in May 2019 that the risk is “the use of AI to automate existing cultural norms,” in “Are we making artificial intelligence biased?” Sharona Hoffman, JD, wrote in September 2019 that AI in medicine raises “concerns about privacy, discrimination, psychological harm, and the physician-patient relationship,” in “Artificial intelligence in medicine raises legal and ethical concerns.” Betty Rabinowitz, MD, wrote in December 2020 that a 2019 study in Science showed that “Black patients are considerably sicker than white patients at any given risk score,” in “The unintended consequences of population health algorithms.” Alex Siauw wrote in October 2025 that “Most AI models in dermatology are trained mostly on images of lighter skin,” in “The dangerous racial bias in dermatology AI.” Craig Hauben, MPA, wrote in June 2026 that “The eGFR error took twenty years to correct when the formula was sitting in plain sight,” in “What the eGFR race correction teaches us about AI.” Dr. Buga Charles George Kenyi wrote in September 2026 that “The places that stand to benefit most from autonomous AI are often the least able to supervise it,” in “Who is left to supervise AI in low-resource settings?“

On The Podcast by KevinMD: “How algorithmic bias created a mental health crisis,” Ronke Lawal, December 2025; and “How AI can eliminate bias and inefficiency in health care billing,” Tim Wetherill, April 2025.

Questions on this page · All records

What do physicians say about AI in medical education?

Students and physicians on KevinMD agree the tools are already in use and disagree about what they cost: a study aid students reach for first, and a shortcut that may teach retrieval before reasoning. The 2026 posts add a concern the earlier ones lacked, that professional identity forms in supervised responsibility and cannot be downloaded. Posts mentioning education or students have held near 30 percent of posts since 2023.

Kathryn Peper, MD, wrote in March 2019 that “Knowing how artificial intelligence works empowers us to be at the forefront of using it,” in “Knowing how artificial intelligence works empowers clinicians to be at the forefront of using it.” Riya Sood, then a medical student, wrote in June 2024 that the tool’s drawbacks “warrant caution and carefulness when using it as a thinking tool in making medical diagnoses,” in “The growing abuse of ChatGPT in medical education.” J. Wesley Boyd, a psychiatrist, said on The Podcast by KevinMD in June 2024 that he had “used AI to generate arguments for the FDA regulation of lab developed tests, and also against,” in “AI ethics in medical education [PODCAST].” Vijay Rajput, MD, wrote in January 2026 that “Professional identity is not something students download or ask GenAI,” in “AI in medical education: the risk to professional identity formation.” Kathleen Muldoon, a professor and coach, said on The Podcast by KevinMD in June 2026 that “the hidden curriculum is going to speak louder than the syllabus,” in “What’s actually behind medical students using AI [PODCAST].” Ritika Segireddy, then a medical student, wrote in August 2026 that “we may miss the slower but essential process of learning how to arrive at that answer ourselves,” in “Learning medicine with AI before learning to think.”

Questions on this page · All records

What do physicians say should govern AI in medicine?

On governance, physicians on KevinMD say a clinical AI tool should be validated like a drug or device, with an explanation a clinician can contest, and they note that insurers’ automated review already runs at a scale physicians cannot match. Posts mentioning regulation, the FDA, or governance rose from 34 percent of 2024 posts to 64 percent in 2025 and 52 percent in 2026.

Rafael Rolon Rivera, MD, a hospitalist, wrote in August 2025 that “Clinical AI must be validated like any drug or device,” in “Why AI in health care needs the same scrutiny as chemotherapy.” Max Rollwage, a health care executive, said on The Podcast by KevinMD in September 2025 that he had regulated “the only mental health chatbot that is a medical device,” in “Why AI in health care needs stronger testing before clinical use [PODCAST].” Tiffiny Black, DM, MBA, MPA, wrote in December 2025 that “No denial should occur without an accessible explanation that clinicians can understand and contest,” in “AI in prior authorization: the new gatekeeper.” Alan P. Feren, MD, wrote in July 2026 that a signature is not authorship, and “Authorship is knowing, before you sign, whether it’s true,” in “Clinical authorship: the missing principle in medical AI.” Kayvan Haddadan, MD, a pain management physician, wrote in September 2026 that “payers can apply automated retrospective scrutiny at scale and retain recovered amounts,” in “Physician reimbursement cuts collide with AI audits.”

Questions on this page · All records

Who do physicians say should design clinical AI?

On who should design the tools, physicians on KevinMD answer in one voice: clinicians, from the design stage, with continuous audit after go-live, and the authors who build tools say it most sharply. Trust, they argue, is a design property rather than a repair, and governance approved in a boardroom has to reach the exam room.

Tod Stillson, MD, a family physician who has built clinical AI software, wrote in January 2026 that “poorly designed AI will replace relationships, obscure accountability, and optimize care for efficiency instead of outcomes,” in “AI in medicine: Why it won’t replace doctors but will redefine them.” Tod Stillson, MD, wrote in February 2026 that “our profession must shape the technology, not the other way around,” in “AI governance in health care: Why physicians must lead the design.” Harvey Castro, MD, MBA, said on The Podcast by KevinMD in March 2026 that “we must have a data scientist in the loop,” in “Hospitals must establish safety guardrails before deploying AI [PODCAST].” Shiv K. Goel, an internal medicine physician, said on The Podcast by KevinMD in March 2026 that when a deployed tool assigns a diagnosis, “I cannot change it because that AI was already installed in the system,” in “Autonomous AI agents could strip the soul from medicine [PODCAST].” Susan Grant, DNP, RN, wrote in June 2026 that success depends on “how involved and aligned nurses and clinicians are in the design, deployment, and continuous improvement of the tools,” in “Clinician trust in AI is not a one-time milestone.” Matt Hasan, PhD, wrote in September 2026 that trust has become “something we try to repair after a failure rather than something designed into the relationship,” in “Trust in health care AI is a design problem, not a patch.” Amanda Heidemann, MD, wrote in September 2026 that a checklist is not enough, which makes “point-of-care governance as important as enterprise governance,” in “AI governance in health care has to reach the exam room.”

On The Podcast by KevinMD: “Why physicians must lead the design of artificial intelligence in health care,” Tod Stillson, March 2026.

Questions on this page · All records

How has the AI conversation changed since 2006?

A long, thin prologue and a short, dense record: 47 posts from 2006 to 2022, then 470 since. ChatGPT was named in 71 percent of 2023 posts and 27 percent of 2026 posts. 2024 was the documentation year, scribes in 36 percent of posts and burnout in 43 percent. 2025 and 2026 are the accountability years: regulation and responsibility each pass half of posts, and the count doubles from 2025 to 2026.

Term	2006 to 2016 (9)	2017 to 2019 (24)	2020 to 2022 (14)	2023 (89)	2024 (67)	2025 (107)	2026 (207)
ChatGPT, GPT, or OpenAI	0 percent	0 percent	0 percent	71 percent	46 percent	33 percent	30 percent
Scribe or ambient	0 percent	8 percent	7 percent	13 percent	36 percent	28 percent	22 percent
Burnout	0 percent	21 percent	21 percent	26 percent	43 percent	22 percent	21 percent
Hallucination or error	0 percent	8 percent	7 percent	28 percent	36 percent	35 percent	37 percent
Responsible or accountable	22 percent	12 percent	7 percent	34 percent	28 percent	50 percent	54 percent
Replacing physicians	11 percent	4 percent	7 percent	8 percent	7 percent	21 percent	21 percent
Regulation, FDA, or governance	11 percent	12 percent	14 percent	22 percent	34 percent	64 percent	52 percent
Bias or equity	0 percent	17 percent	21 percent	20 percent	31 percent	46 percent	28 percent

Share of posts in each period using the term at least once, on the full text of the 517 posts as of September 20, 2026; parentheses give the post count. The prologue runs from Kevin Pho’s 2006 note on the Goldman chest pain algorithm, counted but not indexed, through Pennell in 2015, Campbell in 2017, the 2018 posts of Plotzker, Skeptical Scalpel, and Kirsch, and the 2019 posts of Peper, Emeli, and Hoffman. The 2024 scribe posts live on the electronic health record page.

Questions on this page · All records

About this record

The KevinMD artificial intelligence corpus by the numbers

The figures below describe the set of KevinMD posts this page draws on, as of September 20, 2026. They are counts of what KevinMD has published, not survey data.

Measure	Value
Posts in the corpus	517
Date range	April 28, 2006 to September 19, 2026
Posts from 2023 onward	470
Posts in 2026 through September 19	207
Named contributors	276
Posts bylined by an MD or DO	298
Podcast episodes	71, 53 with full transcripts
Posts with ChatGPT in the title	34
Posts not indexed, counted but not cited	7
Posts cited on this page	69 of 517, 13 percent of the corpus
Total words	About 586,000
Most frequent contributors	Harvey Castro, MD, MBA (58); Arthur Lazarus, MD, MBA (16); Nicole Drapeau Gillen (9); Matt Hasan, PhD (7); Neil Anand, MD (6)

How this page was built and how it is updated

The corpus was assembled on September 20, 2026 from title searches for “artificial intelligence,” “ChatGPT,” “machine learning,” “algorithm,” “large language model,” and “generative,” plus a scan of all 45,773 KevinMD titles for the whole word “AI,” which cannot be searched directly because it matches every title containing “pain” or “said.” Five posts using “algorithm” in unrelated senses, nine where “generative” matched inside “degenerative” or “regenerative,” and one subscriber spotlight were removed. Term frequencies are computed on the full text of each post from fixed expressions stored with the corpus, so every figure here and on the index can be recomputed; three figures on the previous version of this page (liability 25 to 45 percent, error or hallucination 41 to 65 percent, burnout in 34 percent of posts) could not be reproduced from any stored expression and were replaced. Posts are counted whether or not they are indexed; citations are limited to indexed posts, and the seven unindexed posts are counted but not cited.

Quotations are verbatim from the original posts. Author credentials are as they appeared in the byline; contributors who build, sell, or consult on AI tools are identified. Harvey Castro, MD, MBA, the most frequent contributor, is cited at the same rate as other frequent voices rather than in proportion to his output; contributors whose AI posts are about scribes are cited on the electronic health record page instead. Every source is linked in the sentence that cites it, and the full list is at the end of the page. The complete feed is at the Health IT and AI in Medicine archive. Related records: The electronic health record: what physicians say, in their own words for scribes and documentation, Prior authorization: what physicians say, in their own words for insurers’ use of AI, Medical malpractice: what physicians say, in their own words, Medical errors: what physicians say, in their own words, and Physician burnout: what physicians say, in their own words.

This page is updated quarterly; at each update the date changes, the counts are recomputed, and new claims are added. Nothing is removed unless the original post is removed. An author who believes a quotation on this page misrepresents them can write to Kevin Pho and the page will be corrected. To cite this page: Pho K. Artificial intelligence: what physicians say, in their own words. KevinMD.com. Updated September 20, 2026. https://kevinmd.com/artificial-intelligence

The 69 KevinMD posts cited on this page, in order of publication

• Nathan A. Pennell, MD, PhD. “This is why a computer algorithm cannot ever fully replace a doctor’s judgment.” KevinMD, December 2015.
• Bruce Campbell, MD. “Artificial intelligence in medicine is still a long ways off.” KevinMD, July 2017.
• Richard Plotzker, MD. “Computer algorithms don’t think like doctors.” KevinMD, April 2018.
• Skeptical Scalpel, MD. “Artificial intelligence in medicine: not ready for prime time.” KevinMD, August 2018.
• Michael Kirsch, MD. “Will artificial intelligence ever replace doctors?” KevinMD, September 2018.
• Kathryn Peper, MD. “Knowing how artificial intelligence works empowers clinicians to be at the forefront of using it.” KevinMD, March 2019.
• Iyesatta Massaquoi Emeli, MD. “Are we making artificial intelligence biased?” KevinMD, May 2019.
• Sharona Hoffman, JD. “Artificial intelligence in medicine raises legal and ethical concerns.” KevinMD, September 2019.
• Betty Rabinowitz, MD. “The unintended consequences of population health algorithms.” KevinMD, December 2020.
• Jennifer Lycette, MD. “I’m a medical oncologist. Here’s why AI isn’t going to cure cancer.” KevinMD, April 2023.
• Brett Mollard, MD. “How AI is enhancing patient care and improving radiologists’ lives.” KevinMD, May 2023.
• Hannah Kopelman, DO, MBA. “How artificial intelligence (AI) is transforming dermatology diagnosis and treatment.” KevinMD, July 2023.
• Harvey Castro, MD, MBA. “Lowering hallucination results with ChatGPT.” KevinMD, July 2023.
• Robert Pearl, MD. “How generative AI will upend the doctor-patient relationship.” KevinMD, December 2023.
• L. Joseph Parker, MD. “AI enforcement in health care: Unpacking the DEA’s approach to the opioid epidemic.” KevinMD, January 2024.
• The Podcast by KevinMD, with J. Wesley Boyd. “AI ethics in medical education [PODCAST].” KevinMD, June 2024.
• Riya Sood. “The growing abuse of ChatGPT in medical education.” KevinMD, June 2024.
• Steven Charlap, MD, MBA. “Towards a future for semi-autonomous AI-powered primary care providers.” KevinMD, October 2024.
• Neil Anand, MD. “Navigating Goodhart’s Law dilemma and the future of AI in medicine.” KevinMD, December 2024.
• Arthur Lazarus, MD, MBA. “AI’s cognitive gap: Why human doctors remain irreplaceable.” KevinMD, December 2024.
• Harvey Castro, MD, MBA. “AI agents in health care: Balancing innovation, trust, and evidence-based implementation.” KevinMD, March 2025.
• P. Dileep Kumar, MD, MBA. “AI in health care: the black box of prior authorization.” KevinMD, April 2025.
• Fardad Behzadi, MD. “AI is not a threat to radiologists. It’s a distraction from what truly matters in medicine.” KevinMD, June 2025.
• Harvey Castro, MD, MBA. “AI and humanity in health care: Preserving what makes us human.” KevinMD, June 2025.
• Bhargav Raman, MD, MBA. “AI is already replacing doctors, just not how you think.” KevinMD, July 2025.
• Rafael Rolon Rivera, MD. “Why AI in health care needs the same scrutiny as chemotherapy.” KevinMD, August 2025.
• The Podcast by KevinMD, with Max Rollwage. “Why AI in health care needs stronger testing before clinical use [PODCAST].” KevinMD, September 2025.
• Ara Feinstein, MD, MPH. “The parallel evolution of computer chess and AI in health care: the inevitable journey to embracing cognitive inferiority.” KevinMD, September 2025.
• Vanessa D’Amario, PhD & Vijay Rajput, MD. “Is it time to embrace augmented empathy while using artificial intelligence in health care?” KevinMD, September 2025.
• The Podcast by KevinMD, with Sujay Jadhav. “How AI is transforming health care with real-world data insights [PODCAST].” KevinMD, October 2025.
• Alex Siauw. “The dangerous racial bias in dermatology AI.” KevinMD, October 2025.
• Ronke Lawal, MBA. “AI companions and loneliness.” KevinMD, October 2025.
• Tiffiny Black, DM, MBA, MPA. “AI in prior authorization: the new gatekeeper.” KevinMD, December 2025.
• Tod Stillson, MD. “AI in medicine: Why it won’t replace doctors but will redefine them.” KevinMD, January 2026.
• P. Dileep Kumar, MD, MBA. “Will AI replace primary care physicians?” KevinMD, January 2026.
• Arthur Lazarus, MD, MBA. “FDA loosens AI oversight: What clinicians need to know about the 2026 guidance.” KevinMD, January 2026.
• Vijay Rajput, MD. “AI in medical education: the risk to professional identity formation.” KevinMD, January 2026.
• Tod Stillson, MD. “AI governance in health care: Why physicians must lead the design.” KevinMD, February 2026.
• Gerald Kuo. “The hidden dangers of AI voice assistants in elder care.” KevinMD, March 2026.
• The Podcast by KevinMD, with Harvey Castro, MD, MBA. “Hospitals must establish safety guardrails before deploying AI [PODCAST].” KevinMD, March 2026.
• The Podcast by KevinMD, with Shiv K. Goel. “Autonomous AI agents could strip the soul from medicine [PODCAST].” KevinMD, March 2026.
• Cybil Sierra Stingl, MD and Robert M. Kaplan, PhD. “How generative AI in health care is changing patient expectations.” KevinMD, April 2026.
• The Podcast by KevinMD, with Dr. Robert Wachter. “Gradually, then suddenly: Dr. Robert Wachter on health care’s giant AI leap [PODCAST].” KevinMD, April 2026.
• Ian Hu and Pao Hsuan Huang. “Why accountability in medicine must guide health care AI.” KevinMD, April 2026.
• Dr. Amarachi Amanda Dukor. “Artificial intelligence in radiology and global health equity.” KevinMD, May 2026.
• Muhamad Aly Rifai, MD. “AI therapy chatbots are crossing into impersonation.” KevinMD, May 2026.
• Susan Grant, DNP, RN. “Clinician trust in AI is not a one-time milestone.” KevinMD, June 2026.
• The Podcast by KevinMD, with Richard E. Anderson. “Why AI has outpaced medical malpractice law, and what to do about it [PODCAST].” KevinMD, June 2026.
• Craig Hauben, MPA. “What the eGFR race correction teaches us about AI.” KevinMD, June 2026.
• Ganesh Asaithambi. “When the AI diagnosis arrives before the patient does.” KevinMD, June 2026.
• The Podcast by KevinMD, with Kathleen Muldoon. “What’s actually behind medical students using AI [PODCAST].” KevinMD, June 2026.
• Erin J. Silvertooth, MD. “Clinical AI liability lands on you, not the vendor.” KevinMD, June 2026.
• Pouyan Golshani, MD. “Is AI replacing radiologists? A radiologist disagrees.” KevinMD, June 2026.
• Neil Panchal, DO. “AI tools for physicians: What to use, what to ignore.” KevinMD, June 2026.
• Rao M. Uppu, PhD. “AI fabricated references threaten medical evidence.” KevinMD, July 2026.
• Alan P. Feren, MD. “Clinical authorship: the missing principle in medical AI.” KevinMD, July 2026.
• Robert Pak, MBA and Thomas Pak, MD, PhD. “AI chatbots and suicide: what our tests found.” KevinMD, August 2026.
• Ritika Segireddy. “Learning medicine with AI before learning to think.” KevinMD, August 2026.
• Simon Mathews, MD. “Before you paste your labs into ChatGPT, read this.” KevinMD, August 2026.
• Ronald L. Lindsay, MD. “Medicine disposed of physicians long before it feared AI.” KevinMD, August 2026.
• Teddy A. Teddy, MD. “When AI misses a diagnosis, the cardiologist signs alone.” KevinMD, September 2026.
• Matt Hasan, PhD. “Trust in health care AI is a design problem, not a patch.” KevinMD, September 2026.
• Andrey Espinoza, MD. “The real risk of AI replacing doctors is not accuracy.” KevinMD, September 2026.
• Timothy Lesaca, MD. “Patient trust in AI is not proof the advice is right.” KevinMD, September 2026.
• Nicole Drapeau Gillen. “AI psychosis is not schizophrenia. The difference matters.” KevinMD, September 2026.
• Dr. Buga Charles George Kenyi. “Who is left to supervise AI in low-resource settings?” KevinMD, September 2026.
• Brian Hudes, MD. “Why clinical AI limitations are not a physician problem.” KevinMD, September 2026.
• Amanda Heidemann, MD. “AI governance in health care has to reach the exam room.” KevinMD, September 2026.
• Kayvan Haddadan, MD. “Physician reimbursement cuts collide with AI audits.” KevinMD, September 2026.