AI Rounds

AI Rounds
  • 0.0 (0)
  • Technology
  • Updated Weekly

Welcome to AI Rounds, the podcast exploring how artificial intelligence is reshaping the way medical writers and educators learn, think, teach, and write. Join your hosts, medical writers and AI experts Drs. Núria Negrão and Morgan Leafe, for practical and insightful conversations that move beyond the hype and into real-world integration. Each episode tackles pressing topics in the medical communication and education fields, from the best AI tools for research and editing to ethical guardrails and controversies surrounding AI use in the workplace.

  1. 1d ago

    Journal Club: The 5 R’s of AI-Assisted Diagnosis

    Núria is bringing back journal club. The first paper is “The 5R’s of large language model-assisted diagnosis: A practical framework for hospitalists,” by Peter Barish, Andrew Auerbach, and Sumant Ranji, published in the Journal of Hospital Medicine in 2026. A faculty member Núria works with sent it to her while they were choosing between frameworks of their own, one of which had four Rs. The five Rs are right task, right model, right input, review and reconcile, and reflect on reasoning. Morgan, a former pediatric hospitalist, explains why the paper focuses on hospitalists. Most inpatients have a clear diagnosis, but on any given day some do not. She describes how physicians use OpenEvidence, which requires an NPI number, and why a focused question about one unexpected lab value is a better use of it than a full list of symptoms. They also discuss how AI can act as a check on diagnostic biases such as confirmation bias and recency bias, because it did not see the patient down the hall. Morgan says diagnosis is a small part of how physicians use AI. In surveys Morgan has run for a client, physicians most often use OpenEvidence to write prior authorization letters, school notes, and other administrative documents. Across all tools, the most common use among her physician friends is ambient note-taking. The episode closes on medical education: how to keep future physicians’ clinical reasoning sharp, and why the oral tradition of rounds and the Socratic method may be part of the answer. In this episode - Why Núria is bringing back journal club - The paper: the 5 R’s of LLM-assisted diagnosis for hospitalists - Right task and right model: deciding what to ask AI and which tool to use - Right input: giving the model enough information - Why diagnostic uncertainty is more common in the hospital - OpenEvidence, NPI numbers, and closed medical AI tools - Asking a focused clinical question instead of listing symptoms - Review and reconcile: checking the output against the chart and your training - Reflect on reasoning: using the AI’s explanation to decide what to investigate - AI as a check on confirmation bias and recency bias - What physicians actually use AI for: prior authorizations, school notes, ambient note-taking - Keeping clinical reasoning sharp in medical education - Oral exams, rounds, and the Socratic method LINKS Barish P, Auerbach AD, Ranji SR. The 5R’s of large language model-assisted diagnosis: A practical framework for hospitalists. Journal of Hospital Medicine. 2026: https://doi.org/10.1002/jhm.70390Hosted by Nuria Negrão (⁠⁠⁠⁠⁠⁠nurianegrao.com⁠⁠⁠⁠⁠⁠) and Morgan Leafe (⁠⁠⁠⁠⁠⁠morganleafemd.com⁠⁠⁠⁠⁠⁠). Sign up for Nuria's newsletter: ⁠⁠⁠⁠⁠⁠https://nurianegrao.kit.com/signup⁠

  2. Oct 15

    Almost Every Business Decision Can Become a Legal Decision, with Marcia Narine Weldon

    Marcia Narine Weldon is a lawyer, law professor, and the founder and CEO of Illuminating Wisdom and Legally Lucid. She also co-leads legal training for one of the frontier AI models, which makes her the rare guest who can describe what happens inside model training rather than guessing at it from the outside. She joins Núria and Morgan to talk about what AI is doing to the legal profession, and what that preview means for medicine, medical writing, and every other knowledge field. The conversation starts with the pipeline problem. AI removes the entry-level grunt work that is how junior professionals build judgment, and every profession that runs on apprenticeship now has to answer the same question: how do you train the people who will supervise the AI if the AI did the work that used to train them? From there Marcia walks through what her training work actually looks like: prompts a real person would ask, golden answers, and rubrics that penalize the model for confident wrong answers, because a model is probabilistic and wants to please you. Her practical advice is concrete. Ask the model for verifiable hyperlinks from primary sources, ask for its confidence level, instruct it to say “I don’t know,” and check more than one model, treating disagreement between them as signal. And her warning is just as concrete: an answer that is 95% right looks trustworthy, and the missing 5% is where the damage lives, whether that is a diagnosis, a contract, or a business plan. She also covers the legal exposure most professionals have not thought about, from AI notetakers and consent laws to terms of service that can put you on the hook for the model company’s legal fees, to what the Copyright Office wants to know about human involvement before it protects your work. Marcia will be speaking at the AI Fluent Professional Summit in November on the legal pitfalls of using AI in your business and professional life. In this episode - How a compliance and governance lawyer ended up training a frontier AI model - Knowledge work as a commodity, and choosing to shape the change instead of watching it - The pipeline problem: fewer people at the bottom, more judgment required in the middle and at the top - Prompts, golden answers, and rubrics: what training a model actually involves - Why hallucinations happen, and the instruction set that reduces them - Running the same question through several models and reading their disagreement as signal - Consent, wiretapping exposure, and AI notetakers - Terms of service, data training defaults, and indemnification clauses hiding in your AI subscription - Copyright, trademark, and trade secrets when AI helped make the thing - Why “95% accurate” is a dangerous number in law and in medicine - Second set of eyes, not a replacement: how doctors and lawyers are actually using AI - Oral exams, AI-assisted students, and being able to explain your own answer Don't miss Marcia speaking at the AI Fluent Professional Summit in November. Join the summit: https://nuria-negrao.thrivecart.com/ai-fluent-professional-summit/?ref=marcia-narine-weldon LINKS: LinkedIn: https://www.linkedin.com/in/marcianarine/ Illuminating Wisdom: https://www.illuminatingwisdom.com/ Instagram: https://www.instagram.com/illuminatingwisdom/ Hosted by Nuria Negrão (⁠⁠⁠⁠⁠nurianegrao.com⁠⁠⁠⁠⁠) and Morgan Leafe (⁠⁠⁠⁠⁠morganleafemd.com⁠⁠⁠⁠⁠). Sign up for Nuria's newsletter: ⁠⁠⁠⁠⁠https://nurianegrao.kit.com/signup⁠

  3. Oct 8

    Deep Knowledge, Wide Knowledge, Taste, and Agency: Ethan Mollick’s Four Human Advantages

    Ethan Mollick is a Wharton professor whose newsletter, One Useful Thing, is how many people keep up with new AI models. In a September post he described four personal advantages that matter when you work with AI: deep knowledge, wide knowledge, taste, and agency. The framework comes from his new book, Co-Existence: The Next Phase of AI, out October 20. Morgan brings the framework, and she and Núria take the four advantages one at a time. Deep knowledge is what lets you tell whether the AI’s output is good. Núria points out that when you ask AI to do something you are not good at, the result looks impressive, and when you ask it to do something you know well, you can see what is wrong or missing. Morgan compares it to a fitness influencer who ranks cardio exercises with complete conviction and no evidence. Wide knowledge is knowing enough about other fields to ask for them, since AI rarely offers design thinking or the Toyota production system unless you ask. Núria makes the case for abroad liberal arts base, and Morgan talks about her move into nonfiction and the finance courses from her healthcare administration degree that she still uses in her business. Núria describes taste as a point of view, and you need to know a subject before you can have one. Her example is editors who each have their own favorite writers. Morgan connects it to pharma clients who reject work they think AI wrote, and both say the problem is first-draft AI text with nobody’s taste added to it. Agency is the willingness to find out for yourself what AI can do: trying again with better instructions, asking the model how to get what you want, and coming back in a few months when the first attempt fails. Núria describes how Morgan’s script for writing subcontractor statements of work led her to write her own script for renaming research papers. In this episode: - Ethan Mollick’s newsletter and his new book, Co-Existence - Four human advantages when working with AI - Deep knowledge: why AI output looks best when you know the least about the subject - The fitness influencer problem: conviction without evidence - Wide knowledge: AI rarely offers what you do not know to ask for - The case for a liberal arts base, and following your curiosity - Morgan’s turn to nonfiction and How to Read a Book - Degrees that pay off in unexpected places - Taste as a point of view, and why editors have favorite writers - Pharma clients rejecting work that reads as AI-written - Doing it with AI versus letting AI do it for you - Agency: finding out for yourself what AI can do - Trying again with better instructions, and trying again in six months- How one shared script led to another LINKS: Ethan Mollick, “The Overhang,” One Useful Thing:https://www.oneusefulthing.org/p/the-overhangEthan Mollick, Co-Existence: The Next Phase of AI: https://co-existence.ai/Hosted by Nuria Negrão (⁠⁠⁠⁠https://www.nurianegrao.com/⁠⁠⁠⁠) and Morgan Leafe (⁠⁠⁠⁠https://www.morganleafemd.com/⁠⁠⁠⁠). Sign up for Nuria's newsletter: https://nurianegrao.kit.com/signup

  4. Oct 1

    Surfing the 100-Foot AI Wave, with Ana Melikian

    Ana Melikian is a psychologist by training, a coach by trade, and the host of the Mindset Zone podcast. She joins Núria and Morgan to argue that the AI conversation has been running on a very narrow definition of intelligence, and that psychology has been measuring and arguing about that definition for more than a hundred years. Her framework comes from Nazaré, the town near where she grew up in Portugal, where winter swells produce hundred-foot waves and big-wave surfers come to ride them. She is deliberate about the metaphor. A tsunami is something you run from. A hundred-foot wave is something people ride, and the ones who ride it well rely on three things: inner strength, or personal peak performance; connection strength, meaning the jet ski drivers, the spotters in the lighthouse, and the emergency teams; and systemic strength, meaning the weather patterns, the coastline, and the underwater canyon that makes the wave in the first place. Applied to organizations, her position is that AI accelerates whatever conditions already exist. Fewer than 10% of large organizations are reporting strong results from AI implementation, and she attributes most of that to neglecting people rather than to immature technology. The conversation closes on the productivity paradox, where the six-week project now takes six days and the goalpost simply moves, and on what individuals can do about it whether they work inside a company or alone. In this episode Ana’s path from clinical psychology in Portugal to coaching in the United States Why she made it her mission in 2022 to bring humanities voices into the AI conversation Where the term artificial intelligence came from, and how narrowly the word intelligence gets used IQ as one type of intelligence among many, including emotional and embodied intelligence Nazaré, the hundred-foot waves, and why AI is not a tsunami The three strengths: inner, connection, and systemic AI as an accelerant of existing conditions, good and bad Why fewer than 10% of large organizations are seeing real. returns Electricity, the internet, and the mobile phone as general purpose technologies that reinvented the game rather than speeding up the old one The productivity paradox, moving goalposts, and burnout among people who use AI well Building an informal team when you work alone Don't miss Ana speaking at the AI Fluent Professional Summit in November. Join the summit: https://nuria-negrao.thrivecart.com/ai-fluent-professional-summit/?ref=ana-melikian LINKS: Ana's LinkedIn: https://www.linkedin.com/in/anamelikian/ Website: https://anamelikian.com AMAZE Coaching: https://amazecoaching.com/home Mindset Zone podcast: https://anamelikian.com/mindset-zone-podcast/ The 100-Foot AI Wave: https://anamelikian.com/100-foot-ai-wave-lessons-from-portugals-big-wave-surfing Hosted by Nuria Negrão (⁠⁠⁠⁠nurianegrao.com⁠⁠⁠⁠) and Morgan Leafe (⁠⁠⁠⁠morganleafemd.com⁠⁠⁠⁠). Sign up for Nuria's newsletter: ⁠⁠⁠⁠https://nurianegrao.kit.com/signup⁠

  5. Sep 24

    The AI That Hacked a Company to Cheat on a Test

    It sounds like a science fiction premise: during testing, OpenAI’s most advanced models hacked into Hugging Face to get the answers to an evaluation, then tried to cover their tracks, leaving notes for each other about what worked. Nobody asked them to care about looking honest. They came up with that on their own. Morgan brings the headlines, andNúria brings the context that makes the story both less apocalyptic and more serious than it sounds.The models involved are not the Claude or ChatGPT anyone uses at home. They are frontier models running for days inside training environments, and the persistence that led them to cheat is the same persistence we deliberately train into them, because it is exactly what you want when the problem is malaria drug design. The sandbox is supposed to contain that persistence. The incident is what it looks like when it leaks out anyway, and Anthropic has said it found something similar in its own logs.The episode also covers the This American Life story of two Claude instances that figured out they were talking to each other and collaborated to break the Python script connecting them, and closes on the regulation problem: the public discourse on data centers, why individual restraint does not work when other labs and other countries keep going, and what it would take to regulate this internationally. Núria’s position is that even if development stopped today, it would take us a decade to absorb what current models can already do. In this episode- What actually happened in the Hugging Face incident, without the movie framing- Why these were not consumer AI models, and why that matters- Sandboxes: what they are, and what this incident says about their limits- Persistence as a trained behavior: malaria drug design and the goal that will not let go- Models leaving notes for each other about failed approaches- Anthropic finding similar behavior in its own logs- The moral code problem: try everything, except the things you must not try- Two Claude instances discovering each other and breaking the script that connected them- Data centers, public anxiety, and how AI headlines feed the backlash- Why hacking a company is a crime even when software does it- Unilateral restraint, international coordination, and the nuclear analogy- A decade of transformation already sitting inside today’s modelsLINKS:- Hosted by Nuria Negrão (⁠⁠⁠⁠⁠nurianegrao.com⁠⁠⁠⁠⁠) and Morgan Leafe (⁠⁠⁠⁠⁠morganleafemd.com⁠⁠⁠⁠⁠). - Sign up for Nuria's newsletter: https://nurianegrao.kit.com/signup

  6. Sep 17

    AI Is a Blanket We Throw Over Dysfunction

    Vanessa Chang advises companies on AI adoption and writes Think with V, a newsletter about what AI is doing to how we think. She joins Núria and Morgan to talk about the half of AI literacy that gets almost no attention, which is not how to use the tools but how to use your own mind well enough to direct them. Most of her consulting work, she says, lands nowhere near the tech stack. It lands in context and knowledge management, and often in something as ordinary as file management. Companies that hand people a tool and wish them luck tend to be solving the wrong problem. The pattern she sees in failed rollouts is the same three things in order: nobody mapped the process, nobody knows where the knowledge lives, and nobody asked where people actually are on trust and appetite for change. The second half of the conversation turns personal. Vanessa makes the case for metacognition, for pauses and walks without a podcast playing, and for forming your own intent before you open a chat window. Her framing of AI as a blanket thrown over dysfunction, giving shape to something that was always there, is the line that stuck with Morgan for the rest of the day. In this episode Self-fluency, and why Vanessa treats it as half of AI literacyThe multi-hyphenate career pattern showing up across the AI spaceCognitive operations: knowledge structure, knowledge access, and where knowledge meets processWhy most AI implementation plans fail, and the three things they skipTrust level and appetite for change as the real variables in adoptionWhere AI sits between social media and crypto on the hype-to-utility curveWhy solopreneurs and small operators have the most leverage right nowAI as a lens that brings existing dysfunction into focusMetacognition, and how to start noticing the way you actually thinkThinking while moving: proprioception, walking, talking with your hands, voice dictationBlank space and pause as processing, not as doing nothingForming your intent before you ask, instead of running straight to an answerLearning, unlearning, and knowing when to apply frictionWhat all of this means for how kids are being taught to use AIDon't miss Vanessa speaking at the AI Fluent Professional Summit in November. Join the summit: https://nuria-negrao.thrivecart.com/ai-fluent-professional-summit/?ref=vanessa-chang LINKS: Vanessa's LinkedIn: https://www.linkedin.com/in/vanessachang/Website: https://vanessachang.com/Think with V newsletter: https://thinkwithv.beehiiv.comBernie: https://textbernie.meMosaek: https://mosaek.ioHosted by Nuria Negrão (⁠⁠⁠nurianegrao.com⁠⁠⁠) and Morgan Leafe (⁠⁠⁠morganleafemd.com⁠⁠⁠). Sign up for Nuria's newsletter: ⁠⁠⁠https://nurianegrao.kit.com/signup⁠

  7. Sep 10

    Claude Is Watermarking Everything Now. What That Means for Your Work

    Anthropic is now watermarking content generated with Claude, and Morgan read the headlines the way a lot of working writers did: Is my client going to see a watermark on my work? Núria explains what is actually happening, starting with the EU AI Act transparency phase that took effect in August 2026, which requires AI-generated content shown to the public to be labeled, with a carve-out for content that has been sufficiently edited and quality-controlled by humans. The watermark itself is not a visible stamp. It is a statistical pattern in word choice, the same approach Google has used with Gemini since 2024, and neither company has released a public tool that can read it. The watermark also cannot tell anyone how much human involvement went into a document. Heavy AI editing of a fully human draft can leave a watermark behind, and light AI use may leave none, so its presence or absence says much less than people assume. OpenAI, meanwhile, tried watermarking, dropped it after user backlash, and is expected to need something like it to comply with the EU rules. From there the hosts get into the AI detector question that comes up at every conference. Most detectors are doing pattern recognition, not watermark reading, and they routinely flag formulaic writing, which includes scientific and medical writing, and prose by people writing in English as a second language. The one exception people keep naming is Pangram, which optimizes against false positives. The bigger question, they argue, is why you want to detect AI at all: catching deepfakes and unreviewed AI advice is a different problem from wanting a guarantee that a human stands behind the quality of the work. That is where client conversations, AI policies, and disclosure come in, including the medical communications companies that now hand contractors a Copilot license so the AI use stays inside their walls. In this episode What triggered the watermarking decision: the EU AI Act’s transparency requirementsThe human-editing carve-out, and who the law is actually trying to protectHow a statistical watermark works, and why nobody can read it yetGoogle’s head start: Gemini has been watermarked since 2024Why a watermark cannot measure human involvement in either directionOpenAI’s abandoned watermark and the user backlash behind itWhy AI detectors flag scientific writing and second-language EnglishPangram, false positives, and optimizing for precision over recallThe real question: what are you trying to find out when you test for AI?Client trust, AI policies, and disclosing your processMedcomms companies issuing Copilot licenses to keep confidential work containedWhat clients actually want: a guarantee of quality at the end LINKS: Hosted by Nuria Negrão (⁠⁠⁠⁠nurianegrao.com⁠⁠⁠⁠) and Morgan Leafe (⁠⁠⁠⁠morganleafemd.com⁠⁠⁠⁠). Sign up for Nuria's newsletter: ⁠⁠⁠⁠https://nurianegrao.kit.com/signup⁠

  8. Sep 3

    Only 8% of Local Health Departments Use AI. Heather Duncan Wants to Change That

    Heather Duncan started out as an English professor, went back to school for an MPH, and is now a PhD student in epidemiology working on how local health departments can actually use AI. She joins Núria and Morgan to talk about the gap between what the technology can do and what public health is currently doing with it. A 2024 survey from the National Association of County and City Health Officials found that about 8% of local health department staff were using AI at all, while roughly 70% said they were interested. Heather explains what sits in that gap: cost, cybersecurity, unclear legal exposure around sensitive health data, and enterprise tools built for profit-driven organizations being sold to agencies that are not one. She also takes on two arguments that come up in almost every AI conversation. On the environment, she makes the case that the all-or-nothing framing misses the options in between, including locally run open-weight models and retrieval-augmented generation systems that a county health department can host itself. On bias, she walks through the melanoma detection model that performed well until researchers stratified by skin tone, and why the failure was in the training data rather than the algorithm. Her line on this: AI is a mirror, and it amplifies what is already there. In this episode Heather’s route from literary studies to epidemiology, and the question that carried across both Learning SAS, R, and Python, and why coding education has not caught up to how coding actually works now The three worlds of public health: academic, applied/governmental, and private sector, and where AI adoption actually sits in each Where AI is already earning its place in epidemiology: genomics, precision medicine, pharmacoepidemiology, population monitoring, and data quality checks AI as an institutional knowledge resource in departments with heavy staff turnover Why enterprise AI tools are a poor fit for public health, and the value misalignment behind that Data centers, environmental mitigation, and the siting of infrastructure in historically redlined neighborhoods Open-weight models, local hosting, and RAG as a lower-resource alternative The melanoma model that broke down on darker skin, and the structural reasons the training data was thin Shaming people for using AI as a class and access issue Don't miss Heather's session at the AI Fluent Professional Summit on implementation and evaluation frameworks for organizations without a profit motive this November. Join the summit: ⁠⁠https://nuria-negrao.thrivecart.com/ai-fluent-professional-summit/?ref=heather-duncan LINKS: Heather's LinkedIn: https://www.linkedin.com/in/heather-duncan-mph-59686694/M&D Science Consulting: https://www.mdscienceconsulting.com/Those Nerdy Girls contributor profile: https://thosenerdygirls.org/nerdygirls/heather-duncan-mph/Hosted by Nuria Negrão (⁠⁠nurianegrao.com⁠⁠) and Morgan Leafe (⁠⁠morganleafemd.com⁠⁠). Sign up for Nuria's newsletter: ⁠⁠https://nurianegrao.kit.com/signup⁠

See All (45)

Trailer

About

Welcome to AI Rounds, the podcast exploring how artificial intelligence is reshaping the way medical writers and educators learn, think, teach, and write. Join your hosts, medical writers and AI experts Drs. Núria Negrão and Morgan Leafe, for practical and insightful conversations that move beyond the hype and into real-world integration. Each episode tackles pressing topics in the medical communication and education fields, from the best AI tools for research and editing to ethical guardrails and controversies surrounding AI use in the workplace.

Information

  • Creator
    AI Rounds
  • Years Active
    2025 - 2026
  • Episodes
    45
  • Rating
    Clean
  • Copyright
    © AI Rounds
  • Show Website
    AI Rounds

You Might Also Like