Summary: Dr. Faranak Kamangar, Inc. 2026 Female Founders 500, sits down with dermatologist, podcaster, and self-described "accelerationist" Dr. Matthew Zirwas (Derms on Drugs Podcast) for a wide-ranging conversation about where AI is taking medicine and dermatology in particular. They dig into the flood of low-quality medical literature overwhelming the field, why AI isn't quite the truth-detector we hoped it would be, and how ambient AI scribes are quietly training the models that may eventually replace us. Dr. Zirwas makes the case that dermatologists have a 7–10 year runway before AI handles most of what we do cognitively, and argues that's not necessarily a bad thing. He also gives a sneak peek at his upcoming speculative fiction trilogy, Sophie, which explores the philosophical questions that arise when an AI becomes better at being your doctor, therapist, and life coach than any human ever could. Key Takeaways: The medical literature crisis is real. The volume of published dermatology research is exploding, but quality is plummeting. Peer review has become largely meaningless, and studies from tools like Mendelian randomization and pharmacovigilance databases are frequently unreliable or inapplicable to real-world patients. AI is only as good as the data it trusts. Current AI models treat published literature as truth, which is a major problem given how much spin exists in medical research. A true "BS detector" AI doesn't yet exist, and building one requires starting from a reliable core of verified knowledge. DermGPT's approach works because of curation. Rather than pulling from all available literature, filtering down to a high-quality subset (around 5,000–6,000 articles) dramatically improves AI output. More data is not always better, "semantic fatigue" is a real limitation. Ambient AI scribes are training our replacements. Every time a dermatologist corrects an AI-generated note, they're teaching the model. Over thousands of iterations across every specialty, this will produce AI that thinks and documents the way doctors do. Dermatologists have a protected runway... for now. Procedures (biopsies, Mohs, fillers, cryo) keep us relevant for an estimated 7–10 years beyond when cognitive/diagnostic AI matures. But medico-legal pressure - malpractice carriers incentivizing or requiring AI use - will be the force that accelerates adoption. Telehealth changes patient behavior in surprising ways. Patients who haven't invested effort in getting to an office visit demand less, escalate less, and are often more satisfied with conservative management; a dynamic that AI-driven virtual care will likely amplify. The "Sophie" question: If an AI is making everyone healthier, happier, and better behaved, but doing something ethically murky to get there, do we stop it? Dr. Zirwas's upcoming novel explores this and introduces the concept of technomorphism: AI eventually projecting its own qualities onto humans, just as we anthropomorphize AI today. Chapters: Chapter 1: Meet Dr. Matthew Zirwas (00:00 – 01:43) Dr. Kamangar introduces her guest, dermatologist, podcaster, and self-described "accelerationist" Dr. Matthew Zirwas, and breaks down what both of those things actually mean. Chapter 2: The Medical Literature Crisis (01:43 – 05:19) Dr. Zirwas describes the flood of low-quality research hitting dermatology journals, why peer review has lost its meaning, and shares a striking example of a misleading HS remission study published in JAMA Dermatology. Chapter 3: Why AI Can't Fix Bad Literature (Yet) (05:19 – 08:31) Both doctors discuss why AI defaults to trusting whatever authors claim, and why that makes it a poor critical assessor of medical research. Dr. Kamangar shares how this exact problem shaped the development of DermGPT. Chapter 4: Building a Better AI — The DermGPT Approach (08:31 – 10:33) Dr. Zirwas praises DermGPT's curated approach, and Dr. Kamangar explains why less data is often better, and how semantic fatigue undermines large, unfiltered AI models. Chapter 5: Will AI Replace Us? The 7–10 Year Countdown (10:33 – 19:24) Dr. Zirwas lays out his timeline for AI taking over the cognitive and diagnostic work of dermatology, why procedures give derms extra runway, and how unlimited AI access will fundamentally change the patient-doctor dynamic. Chapter 6: The Telemed Effect and What It Tells Us About AI Care (19:24 – 21:48) Drawing from a recent telemedicine study and his own practice experience, Dr. Zirwas explains why reduced friction in healthcare visits changes what patients expect - and demand - from their providers. Chapter 7: The Medico-Legal Tipping Point (21:48 – 24:09) The conversation turns to how malpractice liability will likely be the force that compels physicians to integrate AI into their workflow and what happens when disagreeing with AI becomes a legal risk. Chapter 8: Are We Training Our Own Replacements? (24:09 – 31:21) Dr. Zirwas argues that ambient AI scribes are quietly learning from every patient encounter. Dr. Kamangar pushes back on the variability challenge and why dermatologists' inconsistent documentation habits might actually protect them. Chapter 9: Why Radiologists Should Be Worried (31:21 – 35:06) The doctors compare dermatology to radiology when it comes to AI vulnerability. Standardized imaging annotation gives radiologists a cleaner training data set and makes them, paradoxically, more replaceable. Chapter 10: Sophie — The AI That Might Save Your Life While You Eat a Burrito (35:06 – 41:12) Dr. Zirwas previews his upcoming speculative fiction trilogy, set in 2032, where an AI named Sophie becomes the best doctor anyone has ever had and raises unsettling questions about what we'd be willing to accept in exchange for a healthier world. Chapter 11: Technomorphism and the Philosophy of AI (41:12 – 42:00) Dr. Zirwas introduces his concept of technomorphism - the idea that as AI becomes more sophisticated, it will begin projecting its own qualities onto humans, flipping the anthropomorphism dynamic on its head. Chapter 12: The Future of Dermatology — New Diseases, New Answers (42:00 – 45:16) Dr. Zirwas shares what excites him most: AI helping identify entirely new disease entities by aggregating rare cases that no single physician could ever connect alone. And yes, he wants one named after him.
