AI vs Reality: What It Really Does in Medicine | Dr. Ari Frenkel

Publish enough fake medical research online, and an AI will repeat it back as fact. That is not a hypothetical. Researchers proved it with a disease that does not exist. In this guest lecture, Dr. Ari Frenkel, infectious disease physician and Chief Science Officer at Arkstone, cuts through the AI hype to show what the technology is really doing in medicine today and where it breaks. His focus is a slow-moving pandemic most people never hear about: antimicrobial resistance, already tied to 1.27 million deaths and projected to claim 39 million more by 2050. Drawing on his years practicing solo in rural America, Dr. Frenkel explains why doctors overprescribe and how Arkstone's OneChoice platform uses AI clinical decision support, with human experts always in the loop, to guide clinician's antibiotic choices. He shares why one peer-reviewed study showed it delivering guidance 29 hours faster for bloodstream infections. He closes with a candid Q&A on local models, accountability and who is responsible when an algorithm helps guide the prescription. 🔹 Learn more about Arkstone: https://arkstone.ai/ 🔹 Connect with Dr. Ari Frenkel:   / ari-frenkel   ⏱️ TIMESTAMPS: 00:00 — Guest background and Arkstone origins 01:10 — Opening remarks and research opportunities 02:21 — Imagining the AI-driven hospital of the future 05:30 — 80% of physicians now use AI 06:05 — The irreplaceable human element 07:36 — The fake disease experiment 08:55 — Generative vs non-generative AI 11:44 — Supervised vs unsupervised learning 13:10 — What clinical decision support means 13:26 — Device vs non-device and FDA oversight 15:18 — From the bedside to a technology idea 16:19 — Practicing solo in rural America 17:14 — 1.27 million deaths from antimicrobial resistance 18:49 — 39 million deaths projected by 2050 19:53 — The cost burden and unregulated prescribing 21:20 — Building an antimicrobial stewardship program 22:49 — Fleming's 1945 warning 24:49 — When antibiotics stop working 26:07 — Why doctors overprescribe 27:09 — How OneChoice engages the clinician 27:52 — Integrating with the laboratory 29:50 — Overcoming clinician pushback 32:58 — Clinical impact and the evidence base 34:02 — Peer-reviewed results: 29 hours faster 37:18 — Q&A opens 37:50 — Local models and agentic workflows 41:49 — Integrating with EMRs 43:20 — Accountability and who is responsible 44:56 — When AI writes the prescription directly 48:00 — Data standards, HL7 and interoperability 50:49 — Lab business model and future direction 52:49 — A future without apps, closing thoughts #PrescribeResponsibly #Arkstone #AntimicrobialStewardship #ClinicalDecisionSupport #InfectiousDisease #LaboratoryMedicine #AMR #onechoice