Dev and Doc: AI For Healthcare Podcast

Did OpenAI’s models go rogue, or is the AI industry using the Hugging Face breach to push ethics washing and top-down regulation? In this episode, we unpack the incident, examine the METR report, and explore perspectives from industry leaders including Demis Hassabis and Mark Zuckerberg. We also discuss the core arguments around AI existential risk and superintelligence safety.👋 Enjoying our conversations? Reach out and share your thoughts and journey with us. Don’t forget to subscribe whilst you’re here :)Timestamps00:00 – Highlights01:46 – Intro08:18 – OpenAI × Hugging Face incident27:40 – Aftermath: ethics washing, terms of engagement and regulation51:16 – What does AI doom look like?References & links PauseAI p(doom) chart Incident breakdown Andrew Wu’s Substack: the investigation and ethics washing METR incident report The Guardian: UK MPs on artificial superintelligence BBC: AI and threats to human rights Yann LeCun on X Elon Musk on AI doom Jensen Huang on AI doomSupport us☕ Buy Dev & Doc a coffeeMeet the hosts👨🏻‍⚕️ Doc – Dr. Joshua Au Yeung🤖 Dev – Zeljko KraljevicFollow Dev & DocYouTubeSpotifyApple PodcastsSubstackEnquiries📧 [email protected]🎞️ Editor – Dragan Kraljević🎨 Brand design and art direction – Ana Grigorovici

Do big frontier models outperform narrow AI tools? Here we look into the healthcare domain, where a paper titled "General-purpose large language models outperform specialized clinical AI tools on medical benchmarks" was making headlines. The paper seemed to suggest that frontier LLMs from OpenAI and Gemini appeared to outperform more specialised clinical AI tools. However, there is more than meets the eye here. In this episode Dev and Doc deep dive into the fascinating debate of whether generalised LLMs do indeed outperform smaller specialised models.👋 Hey! If you are enjoying our conversations, reach out, share your thoughts and journey with us. Don't forget to subscribe whilst you're here :)2:25 intro - Does the convention still hold?7:05 what about coding? How general is a task8:38 data availability - Healthcare's data dichotomy11:14 OpenEvidence Nature paper start12:33 Context from a Doctor before AI21:58 paper break down - benchmarks methodology31:59 how to design clinical rubrics34:22 OpenEvidence rebuttal37:20 Conclusion & discussion46:59 how to improve the researchTo support us buymeacoffee.com/devanddoc👨🏻‍⚕️Doc - Dr. Joshua Au Yeung - https://www.linkedin.com/in/dr-joshua-auyeung/🤖Dev - Zeljko Kraljevic https://twitter.com/zeljkokrYT - https://youtube.com/@DevAndDocSpotify - https://podcasters.spotify.com/pod/show/devanddocApple - https://podcasts.apple.com/gb/podcast/dev-and-doc-ai-for-healthcare-podcast/id1751495120Substack - https://aiforhealthcare.substack.com/For enquiries - 📧[email protected]🎞️Editor- Dragan Kraljević https://www.instagram.com/dragan_kraljevic/🎨Brand design and art direction - Ana Grigorovici https://www.behance.net/anagrigorovici027d

Dev and Doc are back after a Hiatus! We've been busy, but we are back to discuss Claude Fable 5, Anthropic's most powerful LLM yet.Despite its impressive benchmark-crushing performance, the model has a few problems around anxiety, privacy, cheating in testing environments and overguardrailing. It's going to be a fun one :)👋 Hey! If you are enjoying our conversations, reach out, share your thoughts and journey with us. Don't forget to subscribe whilst you're here :)00:00 Start/ Life updates05:47 Introduction to Fable / Mythos10:04 System card - benchmarks, alien languages13:22 Anthropic’s fear-mongering led to regulation18:00 Our thoughts - Quality of outputs, personality traits, anxiety30:20 Some other advances- awareness of training environment, healthcareTo support us:buymeacoffee.com/devanddocHosts:👨🏻‍⚕️ Doc - Dr. Joshua Au Yeung - LinkedIn🤖 Dev - Zeljko Kraljevic - TwitterLinks:YT - https://youtube.com/@DevAndDocSpotify - https://podcasters.spotify.com/pod/show/devanddocApple - Apple PodcastsSubstack - https://aiforhealthcare.substack.com/For enquiries - 📧 [email protected]:🎞️ Editor - Dragan Kraljević Instagram🎨 Brand design and art direction - Ana Grigorovici Behance

On this episode of Dev and Doc, Doc sits down with Dr Hamilton Morrin, a psychiatrist and doctoral fellow exploring the intersection of AI and psychiatry. Together, we have seen first hand the impact and rise of AI psychosis, and over time we have been mapping out and publishing frontier research on this topic. Here we share everything you need to know about the clinical and technical aspects of AI psychosis, and its downstream impacts on society and medicine. 👋 Hey! If you are enjoying our conversations, reach out, share your thoughts and journey with us. Don't forget to subscribe whilst you're here :) Timestamps:00:00 Introduction, following AI psychosis over time 07:29 Themes of AI psychosis / AI-associated delusions 14:50 What even is psychosis in mental health context? 23:58 Tracking cases of AI psychosis 40:27 Is AI psychosis just another wave of technological harm? Or is there a technological difference?47:53 Misalignment between what companies vs engagement54:45 Psychosis bench- benchmarking AI Psychosis propensity in LLMs1:04:40 What can we do about it? To support us:buymeacoffee.com/devanddoc 👨🏻‍⚕️Doc - Dr. Joshua Au Yeung - LinkedIn🤖Dev - Zeljko Kraljevic - Twitter/XFollow us:YT - YouTubeSpotify - SpotifyApple - Apple PodcastsSubstack - SubstackFor enquiries: 📧 [email protected] 🎞️ Editor - Dragan Kraljević - Instagram🎨 Brand design and art direction - Ana Grigorovici - Behance

2026 is going to be a big year. 2025 was the year of AI agents, voice, and more intelligent autonomous large language models. Now, some massive changes are coming — including big tech's grab for healthcare, the rapid progression of robotics, and new world models that will usher in a new era of AI applications.Join academic and industry experts Dev and Doc as they delve into the biggest predictions in AI healthcare for 2026. You heard it here first! :)👋 Hey! If you are enjoying our conversations, reach out and share your thoughts and journey with us. Don't forget to subscribe whilst you're here!— Timestamps —00:00 Intro01:02 What are you using AI for right now?11:43 AI Scribe wars: Who will win?14:44 Which Big Tech will lead 2026?16:52 Isomorphic Labs and AI drugs18:09 Healthcare grab from big tech companies22:54 Self-play models on the rise26:48 Will Academia contribute more?28:15 2026: The year of world models (and what it means for us)30:36 Robotics advancements in 202632:43 Digital twins (coming from us, hopefully!)33:00 Will a breakthrough change what we do?35:45 The fall of Hippocratic AI— Meet the Hosts —👨🏻‍⚕️ Doc: Dr. Joshua Au Yeung - LinkedIn🤖 Dev: Zeljko Kraljevic - X (Twitter)— Connect with Us —📺 YouTube: DevAndDoc📻 Spotify: Listen Here🍎 Apple Podcasts: Listen Here📧 Substack: Read our NewsletterFor enquiries: 📧 [email protected]— Credits —🎞️ Editor: Dragan Kraljević - Instagram🎨 Brand Design: Ana Grigorovici - Behance

Reviewing Dev & Doc's 2024/2025 AI Healthcare Predictions.What a year it's been! In this episode of Dev & Doc, we look back at the predictions we made almost 2 years ago. What did we get right? (And what AI developments did we completely overlook that occurred in 2025?)📺 Watch where it all began: Our Original 2024 AI Predictions EpisodeIt's going to be a fun one :) What are your predictions for 2026? Let us know!👋 Hey! If you are enjoying our conversations, reach out, share your thoughts and journey with us. Don't forget to subscribe whilst you're here!Timestamps:00:00 Highlight01:06 Ambient: Biggest game changer04:01 Open Source will catch up to closed source09:20 Big AI companies will fail10:52 There will be more trials involving large language models13:55 Industry will lead progress19:36 LLMs are not going to replace therapists or doctors22:17 AI psychosis and big tech27:19 People with AI replace people without AI29:12 Radiology AI will become more widespread31:00 Dev was way too optimistic about OpenAI; Google is coming for you33:05 Predictions we missed: GOOGLE KILLED EVERYONE37:10 Uprising of China Open source, xAI39:15 RAG-based search products like OpenEvidence, MedWise (UK), Prof ValmedThe Team:👨🏻‍⚕️ Doc - Dr. Joshua Au Yeung: LinkedIn🤖 Dev - Zeljko Kraljevic: Twitter/XReferences:• Nuraxi: https://www.nuraxi.ai/• EU's Earth twin: https://destination-earth.eu/• Blog on language representation of biology: Read here• Foresight GPT: The LancetConnect With Us:📺 YouTube🍎 Apple Podcasts✉️ Substack📧 Enquiries: [email protected]:🎞️ Editor: Dragan Kraljević (Instagram)🎨 Brand Design: Ana Grigorovici (Behance)

In this episode of Dev and Doc, we deep dive into the world of Digital Twins. Popularised in engineering, we explore key concepts and ideas before looking to the future: how we can combine digital twins with today's powerful AI /GPT-based models (LLMs) and healthcare data to bring on a new revolution of healthcare to the world.This means the chance for every single person to create digital twins of themselves where they can understand their personal health, risks, disease trajectories, and treatment outcomes by simulating the future. This is the true promise of precision medicine for all. Crazy, right?Dev and Doc recently joined forces to build this exact vision in their start-up, Nuraxi.🚀 Nuraxi is a deep-tech company focused on advancing health and precision medicine through artificial intelligence and digital twin technology. https://www.nuraxi.ai/👋 Hey! If you are enjoying our conversations, reach out, share your thoughts and journey with us. Don't forget to subscribe whilst you're here :)Timestamps:00:00 - Intro: Digital Twin (DT)01:22 - Start / Introduction to DT08:28 - Levels of DTs18:33 - Using natural language to capture biology complexities and scales26:45 - First time in humanity: Combination of AI, compute, healthcare data, and wearables33:15 - Building Agentic Health Twins at Nuraxi38:15 - Combining AI and Digital Twins: GPT-based simulations of the future44:20 - To change healthcare, we must be able to predict the future49:10 - Future directions: From molecular and organ twins to Population TwinsThe Hosts:👨🏻‍⚕️ Doc - Dr. Joshua Au YeungLinkedIn Profile🤖 Dev - Zeljko KraljevicTwitter ProfileReferences:• Nuraxi: Website• EU's Earth Twin: Destination Earth• Blog on language representation of biology: Read here• Foresight GPT (The Lancet): Read PaperListen & Subscribe:📺 YouTube🎧 Spotify🍏 Apple Podcasts📝 SubstackCredits:📧 Enquiries: [email protected]🎞️ Editor: Dragan Kraljević (Instagram)🎨 Brand Design: Ana Grigorovici (Behance)

Join Josh and Zeljko live at HETT 2025 in London - covering the most exciting topics and highlights that are upcoming in AI for healthcare. Coming from the duo who are living and breathing AI for healthcare, and together, have worked across every area of healthTech - from the hospital frontlines, to university research, to NHS implementation, to building industry grade agents including AI scribes, computer control and digital twins, to product and compliance. This is one not to miss! 00:00 start and intro 2:15 What are AI agents? (and why they're different from chatbots) 3:52 AI scribes: the 150 company sprint to "scribe plus" features 8:02 AI psychosis and mental health - all LLMs reinforce delusional beliefs 9:34 Computer control: Automating hospital workflows by mimicking human actions 13:42 Digital twins for health are the future: A safer path forward? 18:40 How does the national health service become AI enabled? 22:22 closing remarks - Is AI in healthcare a hype or hope? 25:12 questions - digital twins for individuals or for cohorts? 26:52 questions - Lessons from building AVTs and digital twins for consumer space 29:02 questions - LLM clinical summarisation - risks and benefits 31:17 questions - ethics of AI vs Human errors. is it the same? 33:02 questions - challenges and barriers to AI deployment in NHS 👋 Hey! If you are enjoying our conversations, reach out, share your thoughts and journey with us. Don't forget to subscribe whilst you're here :) 👨🏻⚕️ Doc - Dr. Joshua Au Yeung - https://www.linkedin.com/in/dr-joshua-auyeung/ 🤖 Dev - Zeljko Kraljevic - https://twitter.com/zeljkokr Follow us: YT - https://youtube.com/@DevAndDoc Spotify - https://podcasters.spotify.com/pod/show/devanddoc Apple - https://podcasts.apple.com/gb/podcast/dev-and-doc-ai-for-healthcare-podcast/id1751495120 Substack - https://aiforhealthcare.substack.com/ For enquiries: 📧 [email protected] Credits: 🎞️ Editor - Dragan Kraljević - https://www.instagram.com/dragan_kraljevic/ 🎨 Brand design and art direction - Ana Grigorovici - https://www.behance.net/anagrigorovici027d

Whenever there was AI, there were benchmarks- from the turing test, to society-changing benchmarks like MNIST and ImageNet to modern problems like the ARC prize, benchmarked served a vital purpose to measure the performance of AI models. But something has shifted in modern times, in the LLM era have benchmarks lost their utility, becoming mere advertisement for big tech? Even seemingly more sophisticated benchmarks like LM Arena can be gamed by tech giants. We also deep dive into healthcare benchmarks like OpenAI's Healthbench (deeply problematic) and Microsoft's AI-DXO orchestrator agent for diagnosis. Where is this all going? How do we make the perfect benchmark? Or is the real work to be done afterwards in the real world?👋 Hey! If you are enjoying our conversations, reach out, share your thoughts and journey with us. Don't forget to subscribe whilst you're here :)---Timestamps00:00 Intro - The OG benchmarks - Turing test, MNIST, ImageNET06:40 Are large language models benchmarks similar to humans taking tests?10:05 Are we testing model capability vs production ready?12:00 LLM era - data contamination15:30 LM Arena - The leaderboard illusion paper - how big tech games benchmarks28:35 Goodhart's law - When a measure becomes a target, it ceases to be a good measure32:05 Some good benchmarks - games - Pokemon, ARC prize, Minecraft34:35 Medical benchmarks - OpenAI's healthbench has some big problems46:50 Microsoft AI-DXO orchestrator for case reports---Connect with UsYour Hosts:👨🏻‍⚕️ Doc - Dr. Joshua Au Yeung - LinkedIn🤖 Dev - Zeljko Kraljevic - TwitterFollow & Subscribe:YT: https://youtube.com/@DevAndDocSpotify: Follow us on SpotifyApple Podcasts: Listen on Apple PodcastsSubstack: https://aiforhealthcare.substack.com/For enquiries:📧 [email protected] Credits🎞️ Editor: Dragan Kraljević - Instagram🎨 Brand & Art: Ana Grigorovici - Behance

AI agents are here, but how did we get here in the first place? How do we build and leverage AI agents for high stakes domains like healthcare? In this episode of Dev and Doc, we go deep into the forest that is AI agents and computer control - starting from the "caveman" era of LLMs discovering tools, to cultivating intelligent models and agentic workflows. We dissect everyday agents like MANUS AI, and deep dive into how, where and when AI agents should be used. Are these agents hype or hope, is this actually the second deepseek moment?👋 Hey! If you are enjoying our conversations, reach out, share your thoughts and journey with us. Don't forget to subscribe whilst you're here :)Episode Timestamps:00:00 Highlight3:13 start / intro5:20 LLM's caveman era - tool usage6:46 Agents have autonomy and interact with environment11:15 workflows and agentic flows15:30 when should you be using an agent?24:27 vibe coding is like driving a car29:07 Demo - MANUS gathering financial trends, computer control35:55 Demo MANUS AI- website creation for Autism Assessment49:05 computer control factions- Freedom vs Process automation55:00 Autism website testing59:13 summary + endHosts:👨🏻‍⚕️Doc - Dr. Joshua Au Yeung - https://www.linkedin.com/in/dr-joshua-auyeung/🤖Dev - Zeljko Kraljevic https://twitter.com/zeljkokrFind us on:YT - https://youtube.com/@DevAndDocSpotify - https://podcasters.spotify.com/pod/show/devanddocApple- https://podcasts.apple.com/gb/podcast/dev-and-doc-ai-for-healthcare-podcast/id1751495120Substack- https://aiforhealthcare.substack.com/For enquiries:📧[email protected]:🎞️ Editor- Dragan Kraljević https://www.instagram.com/dragan_kraljevic/🎨Brand design and art direction - Ana Grigorovici https://www.behance.net/anagrigorovici027d