• NVIDIA CEO Drops the Blueprint for Europe’s AI Boom

    At GTC Paris — held alongside VivaTech, Europe’s largest tech event — NVIDIA founder and CEO Jensen Huang delivered a clear message: Europe isn’t just adopting AI — it’s building it.
    “We now have a new industry, an AI industry, and it’s now part of the new infrastructure, called intelligence infrastructure, that will be used by every country, every society,” Huang said, addressing an audience gathered online and at the iconic Dôme de Paris.
    From exponential inference growth to quantum breakthroughs, and from infrastructure to industry, agentic AI to robotics, Huang outlined how the region is laying the groundwork for an AI-powered future.

    A New Industrial Revolution
    At the heart of this transformation, Huang explained, are systems like GB200 NVL72 — “one giant GPU” and NVIDIA’s most powerful AI platform yet — now in full production and powering everything from sovereign models to quantum computing.
    “This machine was designed to be a thinking machine, a thinking machine, in the sense that it reasons, it plans, it spends a lot of time talking to itself,” Huang said, walking the audience through the size and scale of these machines and their performance.
    At GTC Paris, Huang showed audience members the innards of some of NVIDIA’s latest hardware.
    There’s more coming, with Huang saying NVIDIA’s partners are now producing 1,000 GB200 systems a week, “and this is just the beginning.” He walked the audience through a range of available systems ranging from the tiny NVIDIA DGX Spark to rack-mounted RTX PRO Servers.
    Huang explained that NVIDIA is working to help countries use technologies like these to build both AI infrastructure — services built for third parties to use and innovate on — and AI factories, which companies build for their own use, to generate revenue.
    NVIDIA is partnering with European governments, telcos and cloud providers to deploy NVIDIA technologies across the region. NVIDIA is also expanding its network of technology centers across Europe — including new hubs in Finland, Germany, Spain, Italy and the U.K. — to accelerate skills development and quantum growth.
    Quantum Meets Classical
    Europe’s quantum ambitions just got a boost.
    The NVIDIA CUDA-Q platform is live on Denmark’s Gefion supercomputer, opening new possibilities for hybrid AI and quantum engineering. In addition, Huang announced that CUDA-Q is now available on NVIDIA Grace Blackwell systems.
    Across the continent, NVIDIA is partnering with supercomputing centers and quantum hardware builders to advance hybrid quantum-AI research and accelerate quantum error correction.
    “Quantum computing is reaching an inflection point,” Huang said. “We are within reach of being able to apply quantum computing, quantum classical computing, in areas that can solve some interesting problems in the coming years.”
    Sovereign Models, Smarter Agents
    European developers want more control over their models. Enter NVIDIA Nemotron, designed to help build large language models tuned to local needs.
    “And so now you know that you have access to an enhanced open model that is still open, that is top of the leader chart,” Huang said.
    These models will be coming to Perplexity, a reasoning search engine, enabling secure, multilingual AI deployment across Europe.
    “You can now ask and get questions answered in the language, in the culture, in the sensibility of your country,” Huang said.
    Huang explained how NVIDIA is helping countries across Europe build AI infrastructure.
    Every company will build its own agents, Huang said. To help create those agents, Huang introduced a suite of agentic AI blueprints, including an Agentic AI Safety blueprint for enterprises and governments.
    The new NVIDIA NeMo Agent toolkit and NVIDIA AI Blueprint for building data flywheels further accelerate the development of safe, high-performing AI agents.
    To help deploy these agents, NVIDIA is partnering with European governments, telcos and cloud providers to deploy the DGX Cloud Lepton platform across the region, providing instant access to accelerated computing capacity.
    “One model architecture, one deployment, and you can run it anywhere,” Huang said, adding that Lepton is now integrated with Hugging Face, giving developers direct access to global compute.
    The Industrial Cloud Goes Live
    AI isn’t just virtual. It’s powering physical systems, too, sparking a new industrial revolution.
    “We’re working on industrial AI with one company after another,” Huang said, describing work to build digital twins based on the NVIDIA Omniverse platform with companies across the continent.
    Huang explained that everything he showed during his keynote was “computer simulation, not animation” and that it looks beautiful because “it turns out the world is beautiful, and it turns out math is beautiful.”
    To further this work, Huang announced NVIDIA is launching the world’s first industrial AI cloud — to be built in Germany — to help Europe’s manufacturers simulate, automate and optimize at scale.
    “Soon, everything that moves will be robotic,” Huang said. “And the car is the next one.”
    NVIDIA DRIVE, NVIDIA’s full-stack AV platform, is now in production to accelerate the large-scale deployment of safe, intelligent transportation.
    And to show what’s coming next, Huang was joined on stage by Grek, a pint-sized robot, as Huang talked about how NVIDIA partnered with DeepMind and Disney to build Newton, the world’s most advanced physics training engine for robotics.
    The Next Wave
    The next wave of AI has begun — and it’s exponential, Huang explained.
    “We have physical robots, and we have information robots. We call them agents,” Huang said. “The technology necessary to teach a robot to manipulate, to simulate — and of course, the manifestation of an incredible robot — is now right in front of us.”
    This new era of AI is being driven by a surge in inference workloads. “The number of people using inference has gone from 8 million to 800 million — 100x in just a couple of years,” Huang said.
    To meet this demand, Huang emphasized the need for a new kind of computer: “We need a special computer designed for thinking, designed for reasoning. And that’s what Blackwell is — a thinking machine.”
    Huang and Grek, as he explained how AI is driving advancements in robotics.
    These Blackwell-powered systems will live in a new class of data centers — AI factories — built to generate tokens, the raw material of modern intelligence.
    “These AI factories are going to generate tokens,” Huang said, turning to Grek with a smile. “And these tokens are going to become your food, little Grek.”
    With that, the keynote closed on a bold vision: a future powered by sovereign infrastructure, agentic AI, robotics — and exponential inference — all built in partnership with Europe.
    Watch the NVIDIA GTC Paris keynote from Huang at VivaTech and explore GTC Paris sessions.
    #nvidia #ceo #drops #blueprint #europes
    NVIDIA CEO Drops the Blueprint for Europe’s AI Boom
    At GTC Paris — held alongside VivaTech, Europe’s largest tech event — NVIDIA founder and CEO Jensen Huang delivered a clear message: Europe isn’t just adopting AI — it’s building it. “We now have a new industry, an AI industry, and it’s now part of the new infrastructure, called intelligence infrastructure, that will be used by every country, every society,” Huang said, addressing an audience gathered online and at the iconic Dôme de Paris. From exponential inference growth to quantum breakthroughs, and from infrastructure to industry, agentic AI to robotics, Huang outlined how the region is laying the groundwork for an AI-powered future. A New Industrial Revolution At the heart of this transformation, Huang explained, are systems like GB200 NVL72 — “one giant GPU” and NVIDIA’s most powerful AI platform yet — now in full production and powering everything from sovereign models to quantum computing. “This machine was designed to be a thinking machine, a thinking machine, in the sense that it reasons, it plans, it spends a lot of time talking to itself,” Huang said, walking the audience through the size and scale of these machines and their performance. At GTC Paris, Huang showed audience members the innards of some of NVIDIA’s latest hardware. There’s more coming, with Huang saying NVIDIA’s partners are now producing 1,000 GB200 systems a week, “and this is just the beginning.” He walked the audience through a range of available systems ranging from the tiny NVIDIA DGX Spark to rack-mounted RTX PRO Servers. Huang explained that NVIDIA is working to help countries use technologies like these to build both AI infrastructure — services built for third parties to use and innovate on — and AI factories, which companies build for their own use, to generate revenue. NVIDIA is partnering with European governments, telcos and cloud providers to deploy NVIDIA technologies across the region. NVIDIA is also expanding its network of technology centers across Europe — including new hubs in Finland, Germany, Spain, Italy and the U.K. — to accelerate skills development and quantum growth. Quantum Meets Classical Europe’s quantum ambitions just got a boost. The NVIDIA CUDA-Q platform is live on Denmark’s Gefion supercomputer, opening new possibilities for hybrid AI and quantum engineering. In addition, Huang announced that CUDA-Q is now available on NVIDIA Grace Blackwell systems. Across the continent, NVIDIA is partnering with supercomputing centers and quantum hardware builders to advance hybrid quantum-AI research and accelerate quantum error correction. “Quantum computing is reaching an inflection point,” Huang said. “We are within reach of being able to apply quantum computing, quantum classical computing, in areas that can solve some interesting problems in the coming years.” Sovereign Models, Smarter Agents European developers want more control over their models. Enter NVIDIA Nemotron, designed to help build large language models tuned to local needs. “And so now you know that you have access to an enhanced open model that is still open, that is top of the leader chart,” Huang said. These models will be coming to Perplexity, a reasoning search engine, enabling secure, multilingual AI deployment across Europe. “You can now ask and get questions answered in the language, in the culture, in the sensibility of your country,” Huang said. Huang explained how NVIDIA is helping countries across Europe build AI infrastructure. Every company will build its own agents, Huang said. To help create those agents, Huang introduced a suite of agentic AI blueprints, including an Agentic AI Safety blueprint for enterprises and governments. The new NVIDIA NeMo Agent toolkit and NVIDIA AI Blueprint for building data flywheels further accelerate the development of safe, high-performing AI agents. To help deploy these agents, NVIDIA is partnering with European governments, telcos and cloud providers to deploy the DGX Cloud Lepton platform across the region, providing instant access to accelerated computing capacity. “One model architecture, one deployment, and you can run it anywhere,” Huang said, adding that Lepton is now integrated with Hugging Face, giving developers direct access to global compute. The Industrial Cloud Goes Live AI isn’t just virtual. It’s powering physical systems, too, sparking a new industrial revolution. “We’re working on industrial AI with one company after another,” Huang said, describing work to build digital twins based on the NVIDIA Omniverse platform with companies across the continent. Huang explained that everything he showed during his keynote was “computer simulation, not animation” and that it looks beautiful because “it turns out the world is beautiful, and it turns out math is beautiful.” To further this work, Huang announced NVIDIA is launching the world’s first industrial AI cloud — to be built in Germany — to help Europe’s manufacturers simulate, automate and optimize at scale. “Soon, everything that moves will be robotic,” Huang said. “And the car is the next one.” NVIDIA DRIVE, NVIDIA’s full-stack AV platform, is now in production to accelerate the large-scale deployment of safe, intelligent transportation. And to show what’s coming next, Huang was joined on stage by Grek, a pint-sized robot, as Huang talked about how NVIDIA partnered with DeepMind and Disney to build Newton, the world’s most advanced physics training engine for robotics. The Next Wave The next wave of AI has begun — and it’s exponential, Huang explained. “We have physical robots, and we have information robots. We call them agents,” Huang said. “The technology necessary to teach a robot to manipulate, to simulate — and of course, the manifestation of an incredible robot — is now right in front of us.” This new era of AI is being driven by a surge in inference workloads. “The number of people using inference has gone from 8 million to 800 million — 100x in just a couple of years,” Huang said. To meet this demand, Huang emphasized the need for a new kind of computer: “We need a special computer designed for thinking, designed for reasoning. And that’s what Blackwell is — a thinking machine.” Huang and Grek, as he explained how AI is driving advancements in robotics. These Blackwell-powered systems will live in a new class of data centers — AI factories — built to generate tokens, the raw material of modern intelligence. “These AI factories are going to generate tokens,” Huang said, turning to Grek with a smile. “And these tokens are going to become your food, little Grek.” With that, the keynote closed on a bold vision: a future powered by sovereign infrastructure, agentic AI, robotics — and exponential inference — all built in partnership with Europe. Watch the NVIDIA GTC Paris keynote from Huang at VivaTech and explore GTC Paris sessions. #nvidia #ceo #drops #blueprint #europes
    BLOGS.NVIDIA.COM
    NVIDIA CEO Drops the Blueprint for Europe’s AI Boom
    At GTC Paris — held alongside VivaTech, Europe’s largest tech event — NVIDIA founder and CEO Jensen Huang delivered a clear message: Europe isn’t just adopting AI — it’s building it. “We now have a new industry, an AI industry, and it’s now part of the new infrastructure, called intelligence infrastructure, that will be used by every country, every society,” Huang said, addressing an audience gathered online and at the iconic Dôme de Paris. From exponential inference growth to quantum breakthroughs, and from infrastructure to industry, agentic AI to robotics, Huang outlined how the region is laying the groundwork for an AI-powered future. A New Industrial Revolution At the heart of this transformation, Huang explained, are systems like GB200 NVL72 — “one giant GPU” and NVIDIA’s most powerful AI platform yet — now in full production and powering everything from sovereign models to quantum computing. “This machine was designed to be a thinking machine, a thinking machine, in the sense that it reasons, it plans, it spends a lot of time talking to itself,” Huang said, walking the audience through the size and scale of these machines and their performance. At GTC Paris, Huang showed audience members the innards of some of NVIDIA’s latest hardware. There’s more coming, with Huang saying NVIDIA’s partners are now producing 1,000 GB200 systems a week, “and this is just the beginning.” He walked the audience through a range of available systems ranging from the tiny NVIDIA DGX Spark to rack-mounted RTX PRO Servers. Huang explained that NVIDIA is working to help countries use technologies like these to build both AI infrastructure — services built for third parties to use and innovate on — and AI factories, which companies build for their own use, to generate revenue. NVIDIA is partnering with European governments, telcos and cloud providers to deploy NVIDIA technologies across the region. NVIDIA is also expanding its network of technology centers across Europe — including new hubs in Finland, Germany, Spain, Italy and the U.K. — to accelerate skills development and quantum growth. Quantum Meets Classical Europe’s quantum ambitions just got a boost. The NVIDIA CUDA-Q platform is live on Denmark’s Gefion supercomputer, opening new possibilities for hybrid AI and quantum engineering. In addition, Huang announced that CUDA-Q is now available on NVIDIA Grace Blackwell systems. Across the continent, NVIDIA is partnering with supercomputing centers and quantum hardware builders to advance hybrid quantum-AI research and accelerate quantum error correction. “Quantum computing is reaching an inflection point,” Huang said. “We are within reach of being able to apply quantum computing, quantum classical computing, in areas that can solve some interesting problems in the coming years.” Sovereign Models, Smarter Agents European developers want more control over their models. Enter NVIDIA Nemotron, designed to help build large language models tuned to local needs. “And so now you know that you have access to an enhanced open model that is still open, that is top of the leader chart,” Huang said. These models will be coming to Perplexity, a reasoning search engine, enabling secure, multilingual AI deployment across Europe. “You can now ask and get questions answered in the language, in the culture, in the sensibility of your country,” Huang said. Huang explained how NVIDIA is helping countries across Europe build AI infrastructure. Every company will build its own agents, Huang said. To help create those agents, Huang introduced a suite of agentic AI blueprints, including an Agentic AI Safety blueprint for enterprises and governments. The new NVIDIA NeMo Agent toolkit and NVIDIA AI Blueprint for building data flywheels further accelerate the development of safe, high-performing AI agents. To help deploy these agents, NVIDIA is partnering with European governments, telcos and cloud providers to deploy the DGX Cloud Lepton platform across the region, providing instant access to accelerated computing capacity. “One model architecture, one deployment, and you can run it anywhere,” Huang said, adding that Lepton is now integrated with Hugging Face, giving developers direct access to global compute. The Industrial Cloud Goes Live AI isn’t just virtual. It’s powering physical systems, too, sparking a new industrial revolution. “We’re working on industrial AI with one company after another,” Huang said, describing work to build digital twins based on the NVIDIA Omniverse platform with companies across the continent. Huang explained that everything he showed during his keynote was “computer simulation, not animation” and that it looks beautiful because “it turns out the world is beautiful, and it turns out math is beautiful.” To further this work, Huang announced NVIDIA is launching the world’s first industrial AI cloud — to be built in Germany — to help Europe’s manufacturers simulate, automate and optimize at scale. “Soon, everything that moves will be robotic,” Huang said. “And the car is the next one.” NVIDIA DRIVE, NVIDIA’s full-stack AV platform, is now in production to accelerate the large-scale deployment of safe, intelligent transportation. And to show what’s coming next, Huang was joined on stage by Grek, a pint-sized robot, as Huang talked about how NVIDIA partnered with DeepMind and Disney to build Newton, the world’s most advanced physics training engine for robotics. The Next Wave The next wave of AI has begun — and it’s exponential, Huang explained. “We have physical robots, and we have information robots. We call them agents,” Huang said. “The technology necessary to teach a robot to manipulate, to simulate — and of course, the manifestation of an incredible robot — is now right in front of us.” This new era of AI is being driven by a surge in inference workloads. “The number of people using inference has gone from 8 million to 800 million — 100x in just a couple of years,” Huang said. To meet this demand, Huang emphasized the need for a new kind of computer: “We need a special computer designed for thinking, designed for reasoning. And that’s what Blackwell is — a thinking machine.” Huang and Grek, as he explained how AI is driving advancements in robotics. These Blackwell-powered systems will live in a new class of data centers — AI factories — built to generate tokens, the raw material of modern intelligence. “These AI factories are going to generate tokens,” Huang said, turning to Grek with a smile. “And these tokens are going to become your food, little Grek.” With that, the keynote closed on a bold vision: a future powered by sovereign infrastructure, agentic AI, robotics — and exponential inference — all built in partnership with Europe. Watch the NVIDIA GTC Paris keynote from Huang at VivaTech and explore GTC Paris sessions.
    Like
    Love
    Sad
    23
    0 Commentarii 0 Distribuiri
  • How AI is reshaping the future of healthcare and medical research

    Transcript       
    PETER LEE: “In ‘The Little Black Bag,’ a classic science fiction story, a high-tech doctor’s kit of the future is accidentally transported back to the 1950s, into the shaky hands of a washed-up, alcoholic doctor. The ultimate medical tool, it redeems the doctor wielding it, allowing him to practice gratifyingly heroic medicine. … The tale ends badly for the doctor and his treacherous assistant, but it offered a picture of how advanced technology could transform medicine—powerful when it was written nearly 75 years ago and still so today. What would be the Al equivalent of that little black bag? At this moment when new capabilities are emerging, how do we imagine them into medicine?”          
    This is The AI Revolution in Medicine, Revisited. I’m your host, Peter Lee.   
    Shortly after OpenAI’s GPT-4 was publicly released, Carey Goldberg, Dr. Zak Kohane, and I published The AI Revolution in Medicine to help educate the world of healthcare and medical research about the transformative impact this new generative AI technology could have. But because we wrote the book when GPT-4 was still a secret, we had to speculate. Now, two years later, what did we get right, and what did we get wrong?    
    In this series, we’ll talk to clinicians, patients, hospital administrators, and others to understand the reality of AI in the field and where we go from here.  The book passage I read at the top is from “Chapter 10: The Big Black Bag.” 
    In imagining AI in medicine, Carey, Zak, and I included in our book two fictional accounts. In the first, a medical resident consults GPT-4 on her personal phone as the patient in front of her crashes. Within seconds, it offers an alternate response based on recent literature. In the second account, a 90-year-old woman with several chronic conditions is living independently and receiving near-constant medical support from an AI aide.   
    In our conversations with the guests we’ve spoken to so far, we’ve caught a glimpse of these predicted futures, seeing how clinicians and patients are actually using AI today and how developers are leveraging the technology in the healthcare products and services they’re creating. In fact, that first fictional account isn’t so fictional after all, as most of the doctors in the real world actually appear to be using AI at least occasionally—and sometimes much more than occasionally—to help in their daily clinical work. And as for the second fictional account, which is more of a science fiction account, it seems we are indeed on the verge of a new way of delivering and receiving healthcare, though the future is still very much open. 
    As we continue to examine the current state of AI in healthcare and its potential to transform the field, I’m pleased to welcome Bill Gates and Sébastien Bubeck.  
    Bill may be best known as the co-founder of Microsoft, having created the company with his childhood friend Paul Allen in 1975. He’s now the founder of Breakthrough Energy, which aims to advance clean energy innovation, and TerraPower, a company developing groundbreaking nuclear energy and science technologies. He also chairs the world’s largest philanthropic organization, the Gates Foundation, and focuses on solving a variety of health challenges around the globe and here at home. 
    Sébastien is a research lead at OpenAI. He was previously a distinguished scientist, vice president of AI, and a colleague of mine here at Microsoft, where his work included spearheading the development of the family of small language models known as Phi. While at Microsoft, he also coauthored the discussion-provoking 2023 paper “Sparks of Artificial General Intelligence,” which presented the results of early experiments with GPT-4 conducted by a small team from Microsoft Research.     
    Here’s my conversation with Bill Gates and Sébastien Bubeck. 
    LEE: Bill, welcome. 
    BILL GATES: Thank you. 
    LEE: Seb … 
    SÉBASTIEN BUBECK: Yeah. Hi, hi, Peter. Nice to be here. 
    LEE: You know, one of the things that I’ve been doing just to get the conversation warmed up is to talk about origin stories, and what I mean about origin stories is, you know, what was the first contact that you had with large language models or the concept of generative AI that convinced you or made you think that something really important was happening? 
    And so, Bill, I think I’ve heard the story about, you know, the time when the OpenAI folks—Sam Altman, Greg Brockman, and others—showed you something, but could we hear from you what those early encounters were like and what was going through your mind?  
    GATES: Well, I’d been visiting OpenAI soon after it was created to see things like GPT-2 and to see the little arm they had that was trying to match human manipulation and, you know, looking at their games like Dota that they were trying to get as good as human play. And honestly, I didn’t think the language model stuff they were doing, even when they got to GPT-3, would show the ability to learn, you know, in the same sense that a human reads a biology book and is able to take that knowledge and access it not only to pass a test but also to create new medicines. 
    And so my challenge to them was that if their LLM could get a five on the advanced placement biology test, then I would say, OK, it took biologic knowledge and encoded it in an accessible way and that I didn’t expect them to do that very quickly but it would be profound.  
    And it was only about six months after I challenged them to do that, that an early version of GPT-4 they brought up to a dinner at my house, and in fact, it answered most of the questions that night very well. The one it got totally wrong, we were … because it was so good, we kept thinking, Oh, we must be wrong. It turned out it was a math weaknessthat, you know, we later understood that that was an area of, weirdly, of incredible weakness of those early models. But, you know, that was when I realized, OK, the age of cheap intelligence was at its beginning. 
    LEE: Yeah. So I guess it seems like you had something similar to me in that my first encounters, I actually harbored some skepticism. Is it fair to say you were skeptical before that? 
    GATES: Well, the idea that we’ve figured out how to encode and access knowledge in this very deep sense without even understanding the nature of the encoding, … 
    LEE: Right.  
    GATES: … that is a bit weird.  
    LEE: Yeah. 
    GATES: We have an algorithm that creates the computation, but even say, OK, where is the president’s birthday stored in there? Where is this fact stored in there? The fact that even now when we’re playing around, getting a little bit more sense of it, it’s opaque to us what the semantic encoding is, it’s, kind of, amazing to me. I thought the invention of knowledge storage would be an explicit way of encoding knowledge, not an implicit statistical training. 
    LEE: Yeah, yeah. All right. So, Seb, you know, on this same topic, you know, I got—as we say at Microsoft—I got pulled into the tent. 
    BUBECK: Yes.  
    LEE: Because this was a very secret project. And then, um, I had the opportunity to select a small number of researchers in MSRto join and start investigating this thing seriously. And the first person I pulled in was you. 
    BUBECK: Yeah. 
    LEE: And so what were your first encounters? Because I actually don’t remember what happened then. 
    BUBECK: Oh, I remember it very well.My first encounter with GPT-4 was in a meeting with the two of you, actually. But my kind of first contact, the first moment where I realized that something was happening with generative AI, was before that. And I agree with Bill that I also wasn’t too impressed by GPT-3. 
    I though that it was kind of, you know, very naturally mimicking the web, sort of parroting what was written there in a nice way. Still in a way which seemed very impressive. But it wasn’t really intelligent in any way. But shortly after GPT-3, there was a model before GPT-4 that really shocked me, and this was the first image generation model, DALL-E 1. 
    So that was in 2021. And I will forever remember the press release of OpenAI where they had this prompt of an avocado chair and then you had this image of the avocado chair.And what really shocked me is that clearly the model kind of “understood” what is a chair, what is an avocado, and was able to merge those concepts. 
    So this was really, to me, the first moment where I saw some understanding in those models.  
    LEE: So this was, just to get the timing right, that was before I pulled you into the tent. 
    BUBECK: That was before. That was like a year before. 
    LEE: Right.  
    BUBECK: And now I will tell you how, you know, we went from that moment to the meeting with the two of you and GPT-4. 
    So once I saw this kind of understanding, I thought, OK, fine. It understands concept, but it’s still not able to reason. It cannot—as, you know, Bill was saying—it cannot learn from your document. It cannot reason.  
    So I set out to try to prove that. You know, this is what I was in the business of at the time, trying to prove things in mathematics. So I was trying to prove that basically autoregressive transformers could never reason. So I was trying to prove this. And after a year of work, I had something reasonable to show. And so I had the meeting with the two of you, and I had this example where I wanted to say, there is no way that an LLM is going to be able to do x. 
    And then as soon as I … I don’t know if you remember, Bill. But as soon as I said that, you said, oh, but wait a second. I had, you know, the OpenAI crew at my house recently, and they showed me a new model. Why don’t we ask this new model this question?  
    LEE: Yeah.
    BUBECK: And we did, and it solved it on the spot. And that really, honestly, just changed my life. Like, you know, I had been working for a year trying to say that this was impossible. And just right there, it was shown to be possible.  
    LEE:One of the very first things I got interested in—because I was really thinking a lot about healthcare—was healthcare and medicine. 
    And I don’t know if the two of you remember, but I ended up doing a lot of tests. I ran through, you know, step one and step two of the US Medical Licensing Exam. Did a whole bunch of other things. I wrote this big report. It was, you know, I can’t remember … a couple hundred pages.  
    And I needed to share this with someone. I didn’t … there weren’t too many people I could share it with. So I sent, I think, a copy to you, Bill. Sent a copy to you, Seb.  
    I hardly slept for about a week putting that report together. And, yeah, and I kept working on it. But I was far from alone. I think everyone who was in the tent, so to speak, in those early days was going through something pretty similar. All right. So I think … of course, a lot of what I put in the report also ended up being examples that made it into the book. 
    But the main purpose of this conversation isn’t to reminisce aboutor indulge in those reminiscences but to talk about what’s happening in healthcare and medicine. And, you know, as I said, we wrote this book. We did it very, very quickly. Seb, you helped. Bill, you know, you provided a review and some endorsements. 
    But, you know, honestly, we didn’t know what we were talking about because no one had access to this thing. And so we just made a bunch of guesses. So really, the whole thing I wanted to probe with the two of you is, now with two years of experience out in the world, what, you know, what do we think is happening today? 
    You know, is AI actually having an impact, positive or negative, on healthcare and medicine? And what do we now think is going to happen in the next two years, five years, or 10 years? And so I realize it’s a little bit too abstract to just ask it that way. So let me just try to narrow the discussion and guide us a little bit.  
    Um, the kind of administrative and clerical work, paperwork, around healthcare—and we made a lot of guesses about that—that appears to be going well, but, you know, Bill, I know we’ve discussed that sometimes that you think there ought to be a lot more going on. Do you have a viewpoint on how AI is actually finding its way into reducing paperwork? 
    GATES: Well, I’m stunned … I don’t think there should be a patient-doctor meeting where the AI is not sitting in and both transcribing, offering to help with the paperwork, and even making suggestions, although the doctor will be the one, you know, who makes the final decision about the diagnosis and whatever prescription gets done.  
    It’s so helpful. You know, when that patient goes home and their, you know, son who wants to understand what happened has some questions, that AI should be available to continue that conversation. And the way you can improve that experience and streamline things and, you know, involve the people who advise you. I don’t understand why that’s not more adopted, because there you still have the human in the loop making that final decision. 
    But even for, like, follow-up calls to make sure the patient did things, to understand if they have concerns and knowing when to escalate back to the doctor, the benefit is incredible. And, you know, that thing is ready for prime time. That paradigm is ready for prime time, in my view. 
    LEE: Yeah, there are some good products, but it seems like the number one use right now—and we kind of got this from some of the previous guests in previous episodes—is the use of AI just to respond to emails from patients.Does that make sense to you? 
    BUBECK: Yeah. So maybe I want to second what Bill was saying but maybe take a step back first. You know, two years ago, like, the concept of clinical scribes, which is one of the things that we’re talking about right now, it would have sounded, in fact, it sounded two years ago, borderline dangerous. Because everybody was worried about hallucinations. What happened if you have this AI listening in and then it transcribes, you know, something wrong? 
    Now, two years later, I think it’s mostly working. And in fact, it is not yet, you know, fully adopted. You’re right. But it is in production. It is used, you know, in many, many places. So this rate of progress is astounding because it wasn’t obvious that we would be able to overcome those obstacles of hallucination. It’s not to say that hallucinations are fully solved. In the case of the closed system, they are.  
    Now, I think more generally what’s going on in the background is that there is something that we, that certainly I, underestimated, which is this management overhead. So I think the reason why this is not adopted everywhere is really a training and teaching aspect. People need to be taught, like, those systems, how to interact with them. 
    And one example that I really like, a study that recently appeared where they tried to use ChatGPT for diagnosis and they were comparing doctors without and with ChatGPT. And the amazing thing … so this was a set of cases where the accuracy of the doctors alone was around 75%. ChatGPT alone was 90%. So that’s already kind of mind blowing. But then the kicker is that doctors with ChatGPT was 80%.  
    Intelligence alone is not enough. It’s also how it’s presented, how you interact with it. And ChatGPT, it’s an amazing tool. Obviously, I absolutely love it. But it’s not … you don’t want a doctor to have to type in, you know, prompts and use it that way. 
    It should be, as Bill was saying, kind of running continuously in the background, sending you notifications. And you have to be really careful of the rate at which those notifications are being sent. Because if they are too frequent, then the doctor will learn to ignore them. So you have to … all of those things matter, in fact, at least as much as the level of intelligence of the machine. 
    LEE: One of the things I think about, Bill, in that scenario that you described, doctors do some thinking about the patient when they write the note. So, you know, I’m always a little uncertain whether it’s actually … you know, you wouldn’t necessarily want to fully automate this, I don’t think. Or at least there needs to be some prompt to the doctor to make sure that the doctor puts some thought into what happened in the encounter with the patient. Does that make sense to you at all? 
    GATES: At this stage, you know, I’d still put the onus on the doctor to write the conclusions and the summary and not delegate that. 
    The tradeoffs you make a little bit are somewhat dependent on the situation you’re in. If you’re in Africa,
    So, yes, the doctor’s still going to have to do a lot of work, but just the quality of letting the patient and the people around them interact and ask questions and have things explained, that alone is such a quality improvement. It’s mind blowing.  
    LEE: So since you mentioned, you know, Africa—and, of course, this touches on the mission and some of the priorities of the Gates Foundation and this idea of democratization of access to expert medical care—what’s the most interesting stuff going on right now? Are there people and organizations or technologies that are impressing you or that you’re tracking? 
    GATES: Yeah. So the Gates Foundation has given out a lot of grants to people in Africa doing education, agriculture but more healthcare examples than anything. And the way these things start off, they often start out either being patient-centric in a narrow situation, like, OK, I’m a pregnant woman; talk to me. Or, I have infectious disease symptoms; talk to me. Or they’re connected to a health worker where they’re helping that worker get their job done. And we have lots of pilots out, you know, in both of those cases.  
    The dream would be eventually to have the thing the patient consults be so broad that it’s like having a doctor available who understands the local things.  
    LEE: Right.  
    GATES: We’re not there yet. But over the next two or three years, you know, particularly given the worsening financial constraints against African health systems, where the withdrawal of money has been dramatic, you know, figuring out how to take this—what I sometimes call “free intelligence”—and build a quality health system around that, we will have to be more radical in low-income countries than any rich country is ever going to be.  
    LEE: Also, there’s maybe a different regulatory environment, so some of those things maybe are easier? Because right now, I think the world hasn’t figured out how to and whether to regulate, let’s say, an AI that might give a medical diagnosis or write a prescription for a medication. 
    BUBECK: Yeah. I think one issue with this, and it’s also slowing down the deployment of AI in healthcare more generally, is a lack of proper benchmark. Because, you know, you were mentioning the USMLE, for example. That’s a great test to test human beings and their knowledge of healthcare and medicine. But it’s not a great test to give to an AI. 
    It’s not asking the right questions. So finding what are the right questions to test whether an AI system is ready to give diagnosis in a constrained setting, that’s a very, very important direction, which to my surprise, is not yet accelerating at the rate that I was hoping for. 
    LEE: OK, so that gives me an excuse to get more now into the core AI tech because something I’ve discussed with both of you is this issue of what are the right tests. And you both know the very first test I give to any new spin of an LLM is I present a patient, the results—a mythical patient—the results of my physical exam, my mythical physical exam. Maybe some results of some initial labs. And then I present or propose a differential diagnosis. And if you’re not in medicine, a differential diagnosis you can just think of as a prioritized list of the possible diagnoses that fit with all that data. And in that proposed differential, I always intentionally make two mistakes. 
    I make a textbook technical error in one of the possible elements of the differential diagnosis, and I have an error of omission. And, you know, I just want to know, does the LLM understand what I’m talking about? And all the good ones out there do now. But then I want to know, can it spot the errors? And then most importantly, is it willing to tell me I’m wrong, that I’ve made a mistake?  
    That last piece seems really hard for AI today. And so let me ask you first, Seb, because at the time of this taping, of course, there was a new spin of GPT-4o last week that became overly sycophantic. In other words, it was actually prone in that test of mine not only to not tell me I’m wrong, but it actually praised me for the creativity of my differential.What’s up with that? 
    BUBECK: Yeah, I guess it’s a testament to the fact that training those models is still more of an art than a science. So it’s a difficult job. Just to be clear with the audience, we have rolled back thatversion of GPT-4o, so now we don’t have the sycophant version out there. 
    Yeah, no, it’s a really difficult question. It has to do … as you said, it’s very technical. It has to do with the post-training and how, like, where do you nudge the model? So, you know, there is this very classical by now technique called RLHF, where you push the model in the direction of a certain reward model. So the reward model is just telling the model, you know, what behavior is good, what behavior is bad. 
    But this reward model is itself an LLM, and, you know, Bill was saying at the very beginning of the conversation that we don’t really understand how those LLMs deal with concepts like, you know, where is the capital of France located? Things like that. It is the same thing for this reward model. We don’t know why it says that it prefers one output to another, and whether this is correlated with some sycophancy is, you know, something that we discovered basically just now. That if you push too hard in optimization on this reward model, you will get a sycophant model. 
    So it’s kind of … what I’m trying to say is we became too good at what we were doing, and we ended up, in fact, in a trap of the reward model. 
    LEE: I mean, you do want … it’s a difficult balance because you do want models to follow your desires and … 
    BUBECK: It’s a very difficult, very difficult balance. 
    LEE: So this brings up then the following question for me, which is the extent to which we think we’ll need to have specially trained models for things. So let me start with you, Bill. Do you have a point of view on whether we will need to, you know, quote-unquote take AI models to med school? Have them specially trained? Like, if you were going to deploy something to give medical care in underserved parts of the world, do we need to do something special to create those models? 
    GATES: We certainly need to teach them the African languages and the unique dialects so that the multimedia interactions are very high quality. We certainly need to teach them the disease prevalence and unique disease patterns like, you know, neglected tropical diseases and malaria. So we need to gather a set of facts that somebody trying to go for a US customer base, you know, wouldn’t necessarily have that in there. 
    Those two things are actually very straightforward because the additional training time is small. I’d say for the next few years, we’ll also need to do reinforcement learning about the context of being a doctor and how important certain behaviors are. Humans learn over the course of their life to some degree that, I’m in a different context and the way I behave in terms of being willing to criticize or be nice, you know, how important is it? Who’s here? What’s my relationship to them?  
    Right now, these machines don’t have that broad social experience. And so if you know it’s going to be used for health things, a lot of reinforcement learning of the very best humans in that context would still be valuable. Eventually, the models will, having read all the literature of the world about good doctors, bad doctors, it’ll understand as soon as you say, “I want you to be a doctor diagnosing somebody.” All of the implicit reinforcement that fits that situation, you know, will be there.
    LEE: Yeah.
    GATES: And so I hope three years from now, we don’t have to do that reinforcement learning. But today, for any medical context, you would want a lot of data to reinforce tone, willingness to say things when, you know, there might be something significant at stake. 
    LEE: Yeah. So, you know, something Bill said, kind of, reminds me of another thing that I think we missed, which is, the context also … and the specialization also pertains to different, I guess, what we still call “modes,” although I don’t know if the idea of multimodal is the same as it was two years ago. But, you know, what do you make of all of the hubbub around—in fact, within Microsoft Research, this is a big deal, but I think we’re far from alone—you know, medical images and vision, video, proteins and molecules, cell, you know, cellular data and so on. 
    BUBECK: Yeah. OK. So there is a lot to say to everything … to the last, you know, couple of minutes. Maybe on the specialization aspect, you know, I think there is, hiding behind this, a really fundamental scientific question of whether eventually we have a singular AGIthat kind of knows everything and you can just put, you know, explain your own context and it will just get it and understand everything. 
    That’s one vision. I have to say, I don’t particularly believe in this vision. In fact, we humans are not like that at all. I think, hopefully, we are general intelligences, yet we have to specialize a lot. And, you know, I did myself a lot of RL, reinforcement learning, on mathematics. Like, that’s what I did, you know, spent a lot of time doing that. And I didn’t improve on other aspects. You know, in fact, I probably degraded in other aspects.So it’s … I think it’s an important example to have in mind. 
    LEE: I think I might disagree with you on that, though, because, like, doesn’t a model have to see both good science and bad science in order to be able to gain the ability to discern between the two? 
    BUBECK: Yeah, no, that absolutely. I think there is value in seeing the generality, in having a very broad base. But then you, kind of, specialize on verticals. And this is where also, you know, open-weights model, which we haven’t talked about yet, are really important because they allow you to provide this broad base to everyone. And then you can specialize on top of it. 
    LEE: So we have about three hours of stuff to talk about, but our time is actually running low.
    BUBECK: Yes, yes, yes.  
    LEE: So I think I want … there’s a more provocative question. It’s almost a silly question, but I need to ask it of the two of you, which is, is there a future, you know, where AI replaces doctors or replaces, you know, medical specialties that we have today? So what does the world look like, say, five years from now? 
    GATES: Well, it’s important to distinguish healthcare discovery activity from healthcare delivery activity. We focused mostly on delivery. I think it’s very much within the realm of possibility that the AI is not only accelerating healthcare discovery but substituting for a lot of the roles of, you know, I’m an organic chemist, or I run various types of assays. I can see those, which are, you know, testable-output-type jobs but with still very high value, I can see, you know, some replacement in those areas before the doctor.  
    The doctor, still understanding the human condition and long-term dialogues, you know, they’ve had a lifetime of reinforcement of that, particularly when you get into areas like mental health. So I wouldn’t say in five years, either people will choose to adopt it, but it will be profound that there’ll be this nearly free intelligence that can do follow-up, that can help you, you know, make sure you went through different possibilities. 
    And so I’d say, yes, we’ll have doctors, but I’d say healthcare will be massively transformed in its quality and in efficiency by AI in that time period. 
    LEE: Is there a comparison, useful comparison, say, between doctors and, say, programmers, computer programmers, or doctors and, I don’t know, lawyers? 
    GATES: Programming is another one that has, kind of, a mathematical correctness to it, you know, and so the objective function that you’re trying to reinforce to, as soon as you can understand the state machines, you can have something that’s “checkable”; that’s correct. So I think programming, you know, which is weird to say, that the machine will beat us at most programming tasks before we let it take over roles that have deep empathy, you know, physical presence and social understanding in them. 
    LEE: Yeah. By the way, you know, I fully expect in five years that AI will produce mathematical proofs that are checkable for validity, easily checkable, because they’ll be written in a proof-checking language like Lean or something but will be so complex that no human mathematician can understand them. I expect that to happen.  
    I can imagine in some fields, like cellular biology, we could have the same situation in the future because the molecular pathways, the chemistry, biochemistry of human cells or living cells is as complex as any mathematics, and so it seems possible that we may be in a state where in wet lab, we see, Oh yeah, this actually works, but no one can understand why. 
    BUBECK: Yeah, absolutely. I mean, I think I really agree with Bill’s distinction of the discovery and the delivery, and indeed, the discovery’s when you can check things, and at the end, there is an artifact that you can verify. You know, you can run the protocol in the wet lab and seeproduced what you wanted. So I absolutely agree with that.  
    And in fact, you know, we don’t have to talk five years from now. I don’t know if you know, but just recently, there was a paper that was published on a scientific discovery using o3- mini. So this is really amazing. And, you know, just very quickly, just so people know, it was about this statistical physics model, the frustrated Potts model, which has to do with coloring, and basically, the case of three colors, like, more than two colors was open for a long time, and o3 was able to reduce the case of three colors to two colors.  
    LEE: Yeah. 
    BUBECK: Which is just, like, astounding. And this is not … this is now. This is happening right now. So this is something that I personally didn’t expect it would happen so quickly, and it’s due to those reasoning models.  
    Now, on the delivery side, I would add something more to it for the reason why doctors and, in fact, lawyers and coders will remain for a long time, and it’s because we still don’t understand how those models generalize. Like, at the end of the day, we are not able to tell you when they are confronted with a really new, novel situation, whether they will work or not. 
    Nobody is able to give you that guarantee. And I think until we understand this generalization better, we’re not going to be willing to just let the system in the wild without human supervision. 
    LEE: But don’t human doctors, human specialists … so, for example, a cardiologist sees a patient in a certain way that a nephrologist … 
    BUBECK: Yeah.
    LEE: … or an endocrinologist might not.
    BUBECK: That’s right. But another cardiologist will understand and, kind of, expect a certain level of generalization from their peer. And this, we just don’t have it with AI models. Now, of course, you’re exactly right. That generalization is also hard for humans. Like, if you have a human trained for one task and you put them into another task, then you don’t … you often don’t know.
    LEE: OK. You know, the podcast is focused on what’s happened over the last two years. But now, I’d like one provocative prediction about what you think the world of AI and medicine is going to be at some point in the future. You pick your timeframe. I don’t care if it’s two years or 20 years from now, but, you know, what do you think will be different about AI in medicine in that future than today? 
    BUBECK: Yeah, I think the deployment is going to accelerate soon. Like, we’re really not missing very much. There is this enormous capability overhang. Like, even if progress completely stopped, with current systems, we can do a lot more than what we’re doing right now. So I think this will … this has to be realized, you know, sooner rather than later. 
    And I think it’s probably dependent on these benchmarks and proper evaluation and tying this with regulation. So these are things that take time in human society and for good reason. But now we already are at two years; you know, give it another two years and it should be really …  
    LEE: Will AI prescribe your medicines? Write your prescriptions? 
    BUBECK: I think yes. I think yes. 
    LEE: OK. Bill? 
    GATES: Well, I think the next two years, we’ll have massive pilots, and so the amount of use of the AI, still in a copilot-type mode, you know, we should get millions of patient visits, you know, both in general medicine and in the mental health side, as well. And I think that’s going to build up both the data and the confidence to give the AI some additional autonomy. You know, are you going to let it talk to you at night when you’re panicked about your mental health with some ability to escalate?
    And, you know, I’ve gone so far as to tell politicians with national health systems that if they deploy AI appropriately, that the quality of care, the overload of the doctors, the improvement in the economics will be enough that their voters will be stunned because they just don’t expect this, and, you know, they could be reelectedjust on this one thing of fixing what is a very overloaded and economically challenged health system in these rich countries. 
    You know, my personal role is going to be to make sure that in the poorer countries, there isn’t some lag; in fact, in many cases, that we’ll be more aggressive because, you know, we’re comparing to having no access to doctors at all. And, you know, so I think whether it’s India or Africa, there’ll be lessons that are globally valuable because we need medical intelligence. And, you know, thank god AI is going to provide a lot of that. 
    LEE: Well, on that optimistic note, I think that’s a good way to end. Bill, Seb, really appreciate all of this.  
    I think the most fundamental prediction we made in the book is that AI would actually find its way into the practice of medicine, and I think that that at least has come true, maybe in different ways than we expected, but it’s come true, and I think it’ll only accelerate from here. So thanks again, both of you.  
    GATES: Yeah. Thanks, you guys. 
    BUBECK: Thank you, Peter. Thanks, Bill. 
    LEE: I just always feel such a sense of privilege to have a chance to interact and actually work with people like Bill and Sébastien.   
    With Bill, I’m always amazed at how practically minded he is. He’s really thinking about the nuts and bolts of what AI might be able to do for people, and his thoughts about underserved parts of the world, the idea that we might actually be able to empower people with access to expert medical knowledge, I think is both inspiring and amazing.  
    And then, Seb, Sébastien Bubeck, he’s just absolutely a brilliant mind. He has a really firm grip on the deep mathematics of artificial intelligence and brings that to bear in his research and development work. And where that mathematics takes him isn’t just into the nuts and bolts of algorithms but into philosophical questions about the nature of intelligence.  
    One of the things that Sébastien brought up was the state of evaluation of AI systems. And indeed, he was fairly critical in our conversation. But of course, the world of AI research and development is just moving so fast, and indeed, since we recorded our conversation, OpenAI, in fact, released a new evaluation metric that is directly relevant to medical applications, and that is something called HealthBench. And Microsoft Research also released a new evaluation approach or process called ADeLe.  
    HealthBench and ADeLe are examples of new approaches to evaluating AI models that are less about testing their knowledge and ability to pass multiple-choice exams and instead are evaluation approaches designed to assess how well AI models are able to complete tasks that actually arise every day in typical healthcare or biomedical research settings. These are examples of really important good work that speak to how well AI models work in the real world of healthcare and biomedical research and how well they can collaborate with human beings in those settings. 
    You know, I asked Bill and Seb to make some predictions about the future. You know, my own answer, I expect that we’re going to be able to use AI to change how we diagnose patients, change how we decide treatment options.  
    If you’re a doctor or a nurse and you encounter a patient, you’ll ask questions, do a physical exam, you know, call out for labs just like you do today, but then you’ll be able to engage with AI based on all of that data and just ask, you know, based on all the other people who have gone through the same experience, who have similar data, how were they diagnosed? How were they treated? What were their outcomes? And what does that mean for the patient I have right now? Some people call it the “patients like me” paradigm. And I think that’s going to become real because of AI within our lifetimes. That idea of really grounding the delivery in healthcare and medical practice through data and intelligence, I actually now don’t see any barriers to that future becoming real.  
    I’d like to extend another big thank you to Bill and Sébastien for their time. And to our listeners, as always, it’s a pleasure to have you along for the ride. I hope you’ll join us for our remaining conversations, as well as a second coauthor roundtable with Carey and Zak.  
    Until next time.  
    #how #reshaping #future #healthcare #medical
    How AI is reshaping the future of healthcare and medical research
    Transcript        PETER LEE: “In ‘The Little Black Bag,’ a classic science fiction story, a high-tech doctor’s kit of the future is accidentally transported back to the 1950s, into the shaky hands of a washed-up, alcoholic doctor. The ultimate medical tool, it redeems the doctor wielding it, allowing him to practice gratifyingly heroic medicine. … The tale ends badly for the doctor and his treacherous assistant, but it offered a picture of how advanced technology could transform medicine—powerful when it was written nearly 75 years ago and still so today. What would be the Al equivalent of that little black bag? At this moment when new capabilities are emerging, how do we imagine them into medicine?”           This is The AI Revolution in Medicine, Revisited. I’m your host, Peter Lee.    Shortly after OpenAI’s GPT-4 was publicly released, Carey Goldberg, Dr. Zak Kohane, and I published The AI Revolution in Medicine to help educate the world of healthcare and medical research about the transformative impact this new generative AI technology could have. But because we wrote the book when GPT-4 was still a secret, we had to speculate. Now, two years later, what did we get right, and what did we get wrong?     In this series, we’ll talk to clinicians, patients, hospital administrators, and others to understand the reality of AI in the field and where we go from here.  The book passage I read at the top is from “Chapter 10: The Big Black Bag.”  In imagining AI in medicine, Carey, Zak, and I included in our book two fictional accounts. In the first, a medical resident consults GPT-4 on her personal phone as the patient in front of her crashes. Within seconds, it offers an alternate response based on recent literature. In the second account, a 90-year-old woman with several chronic conditions is living independently and receiving near-constant medical support from an AI aide.    In our conversations with the guests we’ve spoken to so far, we’ve caught a glimpse of these predicted futures, seeing how clinicians and patients are actually using AI today and how developers are leveraging the technology in the healthcare products and services they’re creating. In fact, that first fictional account isn’t so fictional after all, as most of the doctors in the real world actually appear to be using AI at least occasionally—and sometimes much more than occasionally—to help in their daily clinical work. And as for the second fictional account, which is more of a science fiction account, it seems we are indeed on the verge of a new way of delivering and receiving healthcare, though the future is still very much open.  As we continue to examine the current state of AI in healthcare and its potential to transform the field, I’m pleased to welcome Bill Gates and Sébastien Bubeck.   Bill may be best known as the co-founder of Microsoft, having created the company with his childhood friend Paul Allen in 1975. He’s now the founder of Breakthrough Energy, which aims to advance clean energy innovation, and TerraPower, a company developing groundbreaking nuclear energy and science technologies. He also chairs the world’s largest philanthropic organization, the Gates Foundation, and focuses on solving a variety of health challenges around the globe and here at home.  Sébastien is a research lead at OpenAI. He was previously a distinguished scientist, vice president of AI, and a colleague of mine here at Microsoft, where his work included spearheading the development of the family of small language models known as Phi. While at Microsoft, he also coauthored the discussion-provoking 2023 paper “Sparks of Artificial General Intelligence,” which presented the results of early experiments with GPT-4 conducted by a small team from Microsoft Research.      Here’s my conversation with Bill Gates and Sébastien Bubeck.  LEE: Bill, welcome.  BILL GATES: Thank you.  LEE: Seb …  SÉBASTIEN BUBECK: Yeah. Hi, hi, Peter. Nice to be here.  LEE: You know, one of the things that I’ve been doing just to get the conversation warmed up is to talk about origin stories, and what I mean about origin stories is, you know, what was the first contact that you had with large language models or the concept of generative AI that convinced you or made you think that something really important was happening?  And so, Bill, I think I’ve heard the story about, you know, the time when the OpenAI folks—Sam Altman, Greg Brockman, and others—showed you something, but could we hear from you what those early encounters were like and what was going through your mind?   GATES: Well, I’d been visiting OpenAI soon after it was created to see things like GPT-2 and to see the little arm they had that was trying to match human manipulation and, you know, looking at their games like Dota that they were trying to get as good as human play. And honestly, I didn’t think the language model stuff they were doing, even when they got to GPT-3, would show the ability to learn, you know, in the same sense that a human reads a biology book and is able to take that knowledge and access it not only to pass a test but also to create new medicines.  And so my challenge to them was that if their LLM could get a five on the advanced placement biology test, then I would say, OK, it took biologic knowledge and encoded it in an accessible way and that I didn’t expect them to do that very quickly but it would be profound.   And it was only about six months after I challenged them to do that, that an early version of GPT-4 they brought up to a dinner at my house, and in fact, it answered most of the questions that night very well. The one it got totally wrong, we were … because it was so good, we kept thinking, Oh, we must be wrong. It turned out it was a math weaknessthat, you know, we later understood that that was an area of, weirdly, of incredible weakness of those early models. But, you know, that was when I realized, OK, the age of cheap intelligence was at its beginning.  LEE: Yeah. So I guess it seems like you had something similar to me in that my first encounters, I actually harbored some skepticism. Is it fair to say you were skeptical before that?  GATES: Well, the idea that we’ve figured out how to encode and access knowledge in this very deep sense without even understanding the nature of the encoding, …  LEE: Right.   GATES: … that is a bit weird.   LEE: Yeah.  GATES: We have an algorithm that creates the computation, but even say, OK, where is the president’s birthday stored in there? Where is this fact stored in there? The fact that even now when we’re playing around, getting a little bit more sense of it, it’s opaque to us what the semantic encoding is, it’s, kind of, amazing to me. I thought the invention of knowledge storage would be an explicit way of encoding knowledge, not an implicit statistical training.  LEE: Yeah, yeah. All right. So, Seb, you know, on this same topic, you know, I got—as we say at Microsoft—I got pulled into the tent.  BUBECK: Yes.   LEE: Because this was a very secret project. And then, um, I had the opportunity to select a small number of researchers in MSRto join and start investigating this thing seriously. And the first person I pulled in was you.  BUBECK: Yeah.  LEE: And so what were your first encounters? Because I actually don’t remember what happened then.  BUBECK: Oh, I remember it very well.My first encounter with GPT-4 was in a meeting with the two of you, actually. But my kind of first contact, the first moment where I realized that something was happening with generative AI, was before that. And I agree with Bill that I also wasn’t too impressed by GPT-3.  I though that it was kind of, you know, very naturally mimicking the web, sort of parroting what was written there in a nice way. Still in a way which seemed very impressive. But it wasn’t really intelligent in any way. But shortly after GPT-3, there was a model before GPT-4 that really shocked me, and this was the first image generation model, DALL-E 1.  So that was in 2021. And I will forever remember the press release of OpenAI where they had this prompt of an avocado chair and then you had this image of the avocado chair.And what really shocked me is that clearly the model kind of “understood” what is a chair, what is an avocado, and was able to merge those concepts.  So this was really, to me, the first moment where I saw some understanding in those models.   LEE: So this was, just to get the timing right, that was before I pulled you into the tent.  BUBECK: That was before. That was like a year before.  LEE: Right.   BUBECK: And now I will tell you how, you know, we went from that moment to the meeting with the two of you and GPT-4.  So once I saw this kind of understanding, I thought, OK, fine. It understands concept, but it’s still not able to reason. It cannot—as, you know, Bill was saying—it cannot learn from your document. It cannot reason.   So I set out to try to prove that. You know, this is what I was in the business of at the time, trying to prove things in mathematics. So I was trying to prove that basically autoregressive transformers could never reason. So I was trying to prove this. And after a year of work, I had something reasonable to show. And so I had the meeting with the two of you, and I had this example where I wanted to say, there is no way that an LLM is going to be able to do x.  And then as soon as I … I don’t know if you remember, Bill. But as soon as I said that, you said, oh, but wait a second. I had, you know, the OpenAI crew at my house recently, and they showed me a new model. Why don’t we ask this new model this question?   LEE: Yeah. BUBECK: And we did, and it solved it on the spot. And that really, honestly, just changed my life. Like, you know, I had been working for a year trying to say that this was impossible. And just right there, it was shown to be possible.   LEE:One of the very first things I got interested in—because I was really thinking a lot about healthcare—was healthcare and medicine.  And I don’t know if the two of you remember, but I ended up doing a lot of tests. I ran through, you know, step one and step two of the US Medical Licensing Exam. Did a whole bunch of other things. I wrote this big report. It was, you know, I can’t remember … a couple hundred pages.   And I needed to share this with someone. I didn’t … there weren’t too many people I could share it with. So I sent, I think, a copy to you, Bill. Sent a copy to you, Seb.   I hardly slept for about a week putting that report together. And, yeah, and I kept working on it. But I was far from alone. I think everyone who was in the tent, so to speak, in those early days was going through something pretty similar. All right. So I think … of course, a lot of what I put in the report also ended up being examples that made it into the book.  But the main purpose of this conversation isn’t to reminisce aboutor indulge in those reminiscences but to talk about what’s happening in healthcare and medicine. And, you know, as I said, we wrote this book. We did it very, very quickly. Seb, you helped. Bill, you know, you provided a review and some endorsements.  But, you know, honestly, we didn’t know what we were talking about because no one had access to this thing. And so we just made a bunch of guesses. So really, the whole thing I wanted to probe with the two of you is, now with two years of experience out in the world, what, you know, what do we think is happening today?  You know, is AI actually having an impact, positive or negative, on healthcare and medicine? And what do we now think is going to happen in the next two years, five years, or 10 years? And so I realize it’s a little bit too abstract to just ask it that way. So let me just try to narrow the discussion and guide us a little bit.   Um, the kind of administrative and clerical work, paperwork, around healthcare—and we made a lot of guesses about that—that appears to be going well, but, you know, Bill, I know we’ve discussed that sometimes that you think there ought to be a lot more going on. Do you have a viewpoint on how AI is actually finding its way into reducing paperwork?  GATES: Well, I’m stunned … I don’t think there should be a patient-doctor meeting where the AI is not sitting in and both transcribing, offering to help with the paperwork, and even making suggestions, although the doctor will be the one, you know, who makes the final decision about the diagnosis and whatever prescription gets done.   It’s so helpful. You know, when that patient goes home and their, you know, son who wants to understand what happened has some questions, that AI should be available to continue that conversation. And the way you can improve that experience and streamline things and, you know, involve the people who advise you. I don’t understand why that’s not more adopted, because there you still have the human in the loop making that final decision.  But even for, like, follow-up calls to make sure the patient did things, to understand if they have concerns and knowing when to escalate back to the doctor, the benefit is incredible. And, you know, that thing is ready for prime time. That paradigm is ready for prime time, in my view.  LEE: Yeah, there are some good products, but it seems like the number one use right now—and we kind of got this from some of the previous guests in previous episodes—is the use of AI just to respond to emails from patients.Does that make sense to you?  BUBECK: Yeah. So maybe I want to second what Bill was saying but maybe take a step back first. You know, two years ago, like, the concept of clinical scribes, which is one of the things that we’re talking about right now, it would have sounded, in fact, it sounded two years ago, borderline dangerous. Because everybody was worried about hallucinations. What happened if you have this AI listening in and then it transcribes, you know, something wrong?  Now, two years later, I think it’s mostly working. And in fact, it is not yet, you know, fully adopted. You’re right. But it is in production. It is used, you know, in many, many places. So this rate of progress is astounding because it wasn’t obvious that we would be able to overcome those obstacles of hallucination. It’s not to say that hallucinations are fully solved. In the case of the closed system, they are.   Now, I think more generally what’s going on in the background is that there is something that we, that certainly I, underestimated, which is this management overhead. So I think the reason why this is not adopted everywhere is really a training and teaching aspect. People need to be taught, like, those systems, how to interact with them.  And one example that I really like, a study that recently appeared where they tried to use ChatGPT for diagnosis and they were comparing doctors without and with ChatGPT. And the amazing thing … so this was a set of cases where the accuracy of the doctors alone was around 75%. ChatGPT alone was 90%. So that’s already kind of mind blowing. But then the kicker is that doctors with ChatGPT was 80%.   Intelligence alone is not enough. It’s also how it’s presented, how you interact with it. And ChatGPT, it’s an amazing tool. Obviously, I absolutely love it. But it’s not … you don’t want a doctor to have to type in, you know, prompts and use it that way.  It should be, as Bill was saying, kind of running continuously in the background, sending you notifications. And you have to be really careful of the rate at which those notifications are being sent. Because if they are too frequent, then the doctor will learn to ignore them. So you have to … all of those things matter, in fact, at least as much as the level of intelligence of the machine.  LEE: One of the things I think about, Bill, in that scenario that you described, doctors do some thinking about the patient when they write the note. So, you know, I’m always a little uncertain whether it’s actually … you know, you wouldn’t necessarily want to fully automate this, I don’t think. Or at least there needs to be some prompt to the doctor to make sure that the doctor puts some thought into what happened in the encounter with the patient. Does that make sense to you at all?  GATES: At this stage, you know, I’d still put the onus on the doctor to write the conclusions and the summary and not delegate that.  The tradeoffs you make a little bit are somewhat dependent on the situation you’re in. If you’re in Africa, So, yes, the doctor’s still going to have to do a lot of work, but just the quality of letting the patient and the people around them interact and ask questions and have things explained, that alone is such a quality improvement. It’s mind blowing.   LEE: So since you mentioned, you know, Africa—and, of course, this touches on the mission and some of the priorities of the Gates Foundation and this idea of democratization of access to expert medical care—what’s the most interesting stuff going on right now? Are there people and organizations or technologies that are impressing you or that you’re tracking?  GATES: Yeah. So the Gates Foundation has given out a lot of grants to people in Africa doing education, agriculture but more healthcare examples than anything. And the way these things start off, they often start out either being patient-centric in a narrow situation, like, OK, I’m a pregnant woman; talk to me. Or, I have infectious disease symptoms; talk to me. Or they’re connected to a health worker where they’re helping that worker get their job done. And we have lots of pilots out, you know, in both of those cases.   The dream would be eventually to have the thing the patient consults be so broad that it’s like having a doctor available who understands the local things.   LEE: Right.   GATES: We’re not there yet. But over the next two or three years, you know, particularly given the worsening financial constraints against African health systems, where the withdrawal of money has been dramatic, you know, figuring out how to take this—what I sometimes call “free intelligence”—and build a quality health system around that, we will have to be more radical in low-income countries than any rich country is ever going to be.   LEE: Also, there’s maybe a different regulatory environment, so some of those things maybe are easier? Because right now, I think the world hasn’t figured out how to and whether to regulate, let’s say, an AI that might give a medical diagnosis or write a prescription for a medication.  BUBECK: Yeah. I think one issue with this, and it’s also slowing down the deployment of AI in healthcare more generally, is a lack of proper benchmark. Because, you know, you were mentioning the USMLE, for example. That’s a great test to test human beings and their knowledge of healthcare and medicine. But it’s not a great test to give to an AI.  It’s not asking the right questions. So finding what are the right questions to test whether an AI system is ready to give diagnosis in a constrained setting, that’s a very, very important direction, which to my surprise, is not yet accelerating at the rate that I was hoping for.  LEE: OK, so that gives me an excuse to get more now into the core AI tech because something I’ve discussed with both of you is this issue of what are the right tests. And you both know the very first test I give to any new spin of an LLM is I present a patient, the results—a mythical patient—the results of my physical exam, my mythical physical exam. Maybe some results of some initial labs. And then I present or propose a differential diagnosis. And if you’re not in medicine, a differential diagnosis you can just think of as a prioritized list of the possible diagnoses that fit with all that data. And in that proposed differential, I always intentionally make two mistakes.  I make a textbook technical error in one of the possible elements of the differential diagnosis, and I have an error of omission. And, you know, I just want to know, does the LLM understand what I’m talking about? And all the good ones out there do now. But then I want to know, can it spot the errors? And then most importantly, is it willing to tell me I’m wrong, that I’ve made a mistake?   That last piece seems really hard for AI today. And so let me ask you first, Seb, because at the time of this taping, of course, there was a new spin of GPT-4o last week that became overly sycophantic. In other words, it was actually prone in that test of mine not only to not tell me I’m wrong, but it actually praised me for the creativity of my differential.What’s up with that?  BUBECK: Yeah, I guess it’s a testament to the fact that training those models is still more of an art than a science. So it’s a difficult job. Just to be clear with the audience, we have rolled back thatversion of GPT-4o, so now we don’t have the sycophant version out there.  Yeah, no, it’s a really difficult question. It has to do … as you said, it’s very technical. It has to do with the post-training and how, like, where do you nudge the model? So, you know, there is this very classical by now technique called RLHF, where you push the model in the direction of a certain reward model. So the reward model is just telling the model, you know, what behavior is good, what behavior is bad.  But this reward model is itself an LLM, and, you know, Bill was saying at the very beginning of the conversation that we don’t really understand how those LLMs deal with concepts like, you know, where is the capital of France located? Things like that. It is the same thing for this reward model. We don’t know why it says that it prefers one output to another, and whether this is correlated with some sycophancy is, you know, something that we discovered basically just now. That if you push too hard in optimization on this reward model, you will get a sycophant model.  So it’s kind of … what I’m trying to say is we became too good at what we were doing, and we ended up, in fact, in a trap of the reward model.  LEE: I mean, you do want … it’s a difficult balance because you do want models to follow your desires and …  BUBECK: It’s a very difficult, very difficult balance.  LEE: So this brings up then the following question for me, which is the extent to which we think we’ll need to have specially trained models for things. So let me start with you, Bill. Do you have a point of view on whether we will need to, you know, quote-unquote take AI models to med school? Have them specially trained? Like, if you were going to deploy something to give medical care in underserved parts of the world, do we need to do something special to create those models?  GATES: We certainly need to teach them the African languages and the unique dialects so that the multimedia interactions are very high quality. We certainly need to teach them the disease prevalence and unique disease patterns like, you know, neglected tropical diseases and malaria. So we need to gather a set of facts that somebody trying to go for a US customer base, you know, wouldn’t necessarily have that in there.  Those two things are actually very straightforward because the additional training time is small. I’d say for the next few years, we’ll also need to do reinforcement learning about the context of being a doctor and how important certain behaviors are. Humans learn over the course of their life to some degree that, I’m in a different context and the way I behave in terms of being willing to criticize or be nice, you know, how important is it? Who’s here? What’s my relationship to them?   Right now, these machines don’t have that broad social experience. And so if you know it’s going to be used for health things, a lot of reinforcement learning of the very best humans in that context would still be valuable. Eventually, the models will, having read all the literature of the world about good doctors, bad doctors, it’ll understand as soon as you say, “I want you to be a doctor diagnosing somebody.” All of the implicit reinforcement that fits that situation, you know, will be there. LEE: Yeah. GATES: And so I hope three years from now, we don’t have to do that reinforcement learning. But today, for any medical context, you would want a lot of data to reinforce tone, willingness to say things when, you know, there might be something significant at stake.  LEE: Yeah. So, you know, something Bill said, kind of, reminds me of another thing that I think we missed, which is, the context also … and the specialization also pertains to different, I guess, what we still call “modes,” although I don’t know if the idea of multimodal is the same as it was two years ago. But, you know, what do you make of all of the hubbub around—in fact, within Microsoft Research, this is a big deal, but I think we’re far from alone—you know, medical images and vision, video, proteins and molecules, cell, you know, cellular data and so on.  BUBECK: Yeah. OK. So there is a lot to say to everything … to the last, you know, couple of minutes. Maybe on the specialization aspect, you know, I think there is, hiding behind this, a really fundamental scientific question of whether eventually we have a singular AGIthat kind of knows everything and you can just put, you know, explain your own context and it will just get it and understand everything.  That’s one vision. I have to say, I don’t particularly believe in this vision. In fact, we humans are not like that at all. I think, hopefully, we are general intelligences, yet we have to specialize a lot. And, you know, I did myself a lot of RL, reinforcement learning, on mathematics. Like, that’s what I did, you know, spent a lot of time doing that. And I didn’t improve on other aspects. You know, in fact, I probably degraded in other aspects.So it’s … I think it’s an important example to have in mind.  LEE: I think I might disagree with you on that, though, because, like, doesn’t a model have to see both good science and bad science in order to be able to gain the ability to discern between the two?  BUBECK: Yeah, no, that absolutely. I think there is value in seeing the generality, in having a very broad base. But then you, kind of, specialize on verticals. And this is where also, you know, open-weights model, which we haven’t talked about yet, are really important because they allow you to provide this broad base to everyone. And then you can specialize on top of it.  LEE: So we have about three hours of stuff to talk about, but our time is actually running low. BUBECK: Yes, yes, yes.   LEE: So I think I want … there’s a more provocative question. It’s almost a silly question, but I need to ask it of the two of you, which is, is there a future, you know, where AI replaces doctors or replaces, you know, medical specialties that we have today? So what does the world look like, say, five years from now?  GATES: Well, it’s important to distinguish healthcare discovery activity from healthcare delivery activity. We focused mostly on delivery. I think it’s very much within the realm of possibility that the AI is not only accelerating healthcare discovery but substituting for a lot of the roles of, you know, I’m an organic chemist, or I run various types of assays. I can see those, which are, you know, testable-output-type jobs but with still very high value, I can see, you know, some replacement in those areas before the doctor.   The doctor, still understanding the human condition and long-term dialogues, you know, they’ve had a lifetime of reinforcement of that, particularly when you get into areas like mental health. So I wouldn’t say in five years, either people will choose to adopt it, but it will be profound that there’ll be this nearly free intelligence that can do follow-up, that can help you, you know, make sure you went through different possibilities.  And so I’d say, yes, we’ll have doctors, but I’d say healthcare will be massively transformed in its quality and in efficiency by AI in that time period.  LEE: Is there a comparison, useful comparison, say, between doctors and, say, programmers, computer programmers, or doctors and, I don’t know, lawyers?  GATES: Programming is another one that has, kind of, a mathematical correctness to it, you know, and so the objective function that you’re trying to reinforce to, as soon as you can understand the state machines, you can have something that’s “checkable”; that’s correct. So I think programming, you know, which is weird to say, that the machine will beat us at most programming tasks before we let it take over roles that have deep empathy, you know, physical presence and social understanding in them.  LEE: Yeah. By the way, you know, I fully expect in five years that AI will produce mathematical proofs that are checkable for validity, easily checkable, because they’ll be written in a proof-checking language like Lean or something but will be so complex that no human mathematician can understand them. I expect that to happen.   I can imagine in some fields, like cellular biology, we could have the same situation in the future because the molecular pathways, the chemistry, biochemistry of human cells or living cells is as complex as any mathematics, and so it seems possible that we may be in a state where in wet lab, we see, Oh yeah, this actually works, but no one can understand why.  BUBECK: Yeah, absolutely. I mean, I think I really agree with Bill’s distinction of the discovery and the delivery, and indeed, the discovery’s when you can check things, and at the end, there is an artifact that you can verify. You know, you can run the protocol in the wet lab and seeproduced what you wanted. So I absolutely agree with that.   And in fact, you know, we don’t have to talk five years from now. I don’t know if you know, but just recently, there was a paper that was published on a scientific discovery using o3- mini. So this is really amazing. And, you know, just very quickly, just so people know, it was about this statistical physics model, the frustrated Potts model, which has to do with coloring, and basically, the case of three colors, like, more than two colors was open for a long time, and o3 was able to reduce the case of three colors to two colors.   LEE: Yeah.  BUBECK: Which is just, like, astounding. And this is not … this is now. This is happening right now. So this is something that I personally didn’t expect it would happen so quickly, and it’s due to those reasoning models.   Now, on the delivery side, I would add something more to it for the reason why doctors and, in fact, lawyers and coders will remain for a long time, and it’s because we still don’t understand how those models generalize. Like, at the end of the day, we are not able to tell you when they are confronted with a really new, novel situation, whether they will work or not.  Nobody is able to give you that guarantee. And I think until we understand this generalization better, we’re not going to be willing to just let the system in the wild without human supervision.  LEE: But don’t human doctors, human specialists … so, for example, a cardiologist sees a patient in a certain way that a nephrologist …  BUBECK: Yeah. LEE: … or an endocrinologist might not. BUBECK: That’s right. But another cardiologist will understand and, kind of, expect a certain level of generalization from their peer. And this, we just don’t have it with AI models. Now, of course, you’re exactly right. That generalization is also hard for humans. Like, if you have a human trained for one task and you put them into another task, then you don’t … you often don’t know. LEE: OK. You know, the podcast is focused on what’s happened over the last two years. But now, I’d like one provocative prediction about what you think the world of AI and medicine is going to be at some point in the future. You pick your timeframe. I don’t care if it’s two years or 20 years from now, but, you know, what do you think will be different about AI in medicine in that future than today?  BUBECK: Yeah, I think the deployment is going to accelerate soon. Like, we’re really not missing very much. There is this enormous capability overhang. Like, even if progress completely stopped, with current systems, we can do a lot more than what we’re doing right now. So I think this will … this has to be realized, you know, sooner rather than later.  And I think it’s probably dependent on these benchmarks and proper evaluation and tying this with regulation. So these are things that take time in human society and for good reason. But now we already are at two years; you know, give it another two years and it should be really …   LEE: Will AI prescribe your medicines? Write your prescriptions?  BUBECK: I think yes. I think yes.  LEE: OK. Bill?  GATES: Well, I think the next two years, we’ll have massive pilots, and so the amount of use of the AI, still in a copilot-type mode, you know, we should get millions of patient visits, you know, both in general medicine and in the mental health side, as well. And I think that’s going to build up both the data and the confidence to give the AI some additional autonomy. You know, are you going to let it talk to you at night when you’re panicked about your mental health with some ability to escalate? And, you know, I’ve gone so far as to tell politicians with national health systems that if they deploy AI appropriately, that the quality of care, the overload of the doctors, the improvement in the economics will be enough that their voters will be stunned because they just don’t expect this, and, you know, they could be reelectedjust on this one thing of fixing what is a very overloaded and economically challenged health system in these rich countries.  You know, my personal role is going to be to make sure that in the poorer countries, there isn’t some lag; in fact, in many cases, that we’ll be more aggressive because, you know, we’re comparing to having no access to doctors at all. And, you know, so I think whether it’s India or Africa, there’ll be lessons that are globally valuable because we need medical intelligence. And, you know, thank god AI is going to provide a lot of that.  LEE: Well, on that optimistic note, I think that’s a good way to end. Bill, Seb, really appreciate all of this.   I think the most fundamental prediction we made in the book is that AI would actually find its way into the practice of medicine, and I think that that at least has come true, maybe in different ways than we expected, but it’s come true, and I think it’ll only accelerate from here. So thanks again, both of you.   GATES: Yeah. Thanks, you guys.  BUBECK: Thank you, Peter. Thanks, Bill.  LEE: I just always feel such a sense of privilege to have a chance to interact and actually work with people like Bill and Sébastien.    With Bill, I’m always amazed at how practically minded he is. He’s really thinking about the nuts and bolts of what AI might be able to do for people, and his thoughts about underserved parts of the world, the idea that we might actually be able to empower people with access to expert medical knowledge, I think is both inspiring and amazing.   And then, Seb, Sébastien Bubeck, he’s just absolutely a brilliant mind. He has a really firm grip on the deep mathematics of artificial intelligence and brings that to bear in his research and development work. And where that mathematics takes him isn’t just into the nuts and bolts of algorithms but into philosophical questions about the nature of intelligence.   One of the things that Sébastien brought up was the state of evaluation of AI systems. And indeed, he was fairly critical in our conversation. But of course, the world of AI research and development is just moving so fast, and indeed, since we recorded our conversation, OpenAI, in fact, released a new evaluation metric that is directly relevant to medical applications, and that is something called HealthBench. And Microsoft Research also released a new evaluation approach or process called ADeLe.   HealthBench and ADeLe are examples of new approaches to evaluating AI models that are less about testing their knowledge and ability to pass multiple-choice exams and instead are evaluation approaches designed to assess how well AI models are able to complete tasks that actually arise every day in typical healthcare or biomedical research settings. These are examples of really important good work that speak to how well AI models work in the real world of healthcare and biomedical research and how well they can collaborate with human beings in those settings.  You know, I asked Bill and Seb to make some predictions about the future. You know, my own answer, I expect that we’re going to be able to use AI to change how we diagnose patients, change how we decide treatment options.   If you’re a doctor or a nurse and you encounter a patient, you’ll ask questions, do a physical exam, you know, call out for labs just like you do today, but then you’ll be able to engage with AI based on all of that data and just ask, you know, based on all the other people who have gone through the same experience, who have similar data, how were they diagnosed? How were they treated? What were their outcomes? And what does that mean for the patient I have right now? Some people call it the “patients like me” paradigm. And I think that’s going to become real because of AI within our lifetimes. That idea of really grounding the delivery in healthcare and medical practice through data and intelligence, I actually now don’t see any barriers to that future becoming real.   I’d like to extend another big thank you to Bill and Sébastien for their time. And to our listeners, as always, it’s a pleasure to have you along for the ride. I hope you’ll join us for our remaining conversations, as well as a second coauthor roundtable with Carey and Zak.   Until next time.   #how #reshaping #future #healthcare #medical
    WWW.MICROSOFT.COM
    How AI is reshaping the future of healthcare and medical research
    Transcript [MUSIC]      [BOOK PASSAGE]   PETER LEE: “In ‘The Little Black Bag,’ a classic science fiction story, a high-tech doctor’s kit of the future is accidentally transported back to the 1950s, into the shaky hands of a washed-up, alcoholic doctor. The ultimate medical tool, it redeems the doctor wielding it, allowing him to practice gratifyingly heroic medicine. … The tale ends badly for the doctor and his treacherous assistant, but it offered a picture of how advanced technology could transform medicine—powerful when it was written nearly 75 years ago and still so today. What would be the Al equivalent of that little black bag? At this moment when new capabilities are emerging, how do we imagine them into medicine?”   [END OF BOOK PASSAGE]     [THEME MUSIC]     This is The AI Revolution in Medicine, Revisited. I’m your host, Peter Lee.    Shortly after OpenAI’s GPT-4 was publicly released, Carey Goldberg, Dr. Zak Kohane, and I published The AI Revolution in Medicine to help educate the world of healthcare and medical research about the transformative impact this new generative AI technology could have. But because we wrote the book when GPT-4 was still a secret, we had to speculate. Now, two years later, what did we get right, and what did we get wrong?     In this series, we’ll talk to clinicians, patients, hospital administrators, and others to understand the reality of AI in the field and where we go from here.   [THEME MUSIC FADES] The book passage I read at the top is from “Chapter 10: The Big Black Bag.”  In imagining AI in medicine, Carey, Zak, and I included in our book two fictional accounts. In the first, a medical resident consults GPT-4 on her personal phone as the patient in front of her crashes. Within seconds, it offers an alternate response based on recent literature. In the second account, a 90-year-old woman with several chronic conditions is living independently and receiving near-constant medical support from an AI aide.    In our conversations with the guests we’ve spoken to so far, we’ve caught a glimpse of these predicted futures, seeing how clinicians and patients are actually using AI today and how developers are leveraging the technology in the healthcare products and services they’re creating. In fact, that first fictional account isn’t so fictional after all, as most of the doctors in the real world actually appear to be using AI at least occasionally—and sometimes much more than occasionally—to help in their daily clinical work. And as for the second fictional account, which is more of a science fiction account, it seems we are indeed on the verge of a new way of delivering and receiving healthcare, though the future is still very much open.  As we continue to examine the current state of AI in healthcare and its potential to transform the field, I’m pleased to welcome Bill Gates and Sébastien Bubeck.   Bill may be best known as the co-founder of Microsoft, having created the company with his childhood friend Paul Allen in 1975. He’s now the founder of Breakthrough Energy, which aims to advance clean energy innovation, and TerraPower, a company developing groundbreaking nuclear energy and science technologies. He also chairs the world’s largest philanthropic organization, the Gates Foundation, and focuses on solving a variety of health challenges around the globe and here at home.  Sébastien is a research lead at OpenAI. He was previously a distinguished scientist, vice president of AI, and a colleague of mine here at Microsoft, where his work included spearheading the development of the family of small language models known as Phi. While at Microsoft, he also coauthored the discussion-provoking 2023 paper “Sparks of Artificial General Intelligence,” which presented the results of early experiments with GPT-4 conducted by a small team from Microsoft Research.    [TRANSITION MUSIC]   Here’s my conversation with Bill Gates and Sébastien Bubeck.  LEE: Bill, welcome.  BILL GATES: Thank you.  LEE: Seb …  SÉBASTIEN BUBECK: Yeah. Hi, hi, Peter. Nice to be here.  LEE: You know, one of the things that I’ve been doing just to get the conversation warmed up is to talk about origin stories, and what I mean about origin stories is, you know, what was the first contact that you had with large language models or the concept of generative AI that convinced you or made you think that something really important was happening?  And so, Bill, I think I’ve heard the story about, you know, the time when the OpenAI folks—Sam Altman, Greg Brockman, and others—showed you something, but could we hear from you what those early encounters were like and what was going through your mind?   GATES: Well, I’d been visiting OpenAI soon after it was created to see things like GPT-2 and to see the little arm they had that was trying to match human manipulation and, you know, looking at their games like Dota that they were trying to get as good as human play. And honestly, I didn’t think the language model stuff they were doing, even when they got to GPT-3, would show the ability to learn, you know, in the same sense that a human reads a biology book and is able to take that knowledge and access it not only to pass a test but also to create new medicines.  And so my challenge to them was that if their LLM could get a five on the advanced placement biology test, then I would say, OK, it took biologic knowledge and encoded it in an accessible way and that I didn’t expect them to do that very quickly but it would be profound.   And it was only about six months after I challenged them to do that, that an early version of GPT-4 they brought up to a dinner at my house, and in fact, it answered most of the questions that night very well. The one it got totally wrong, we were … because it was so good, we kept thinking, Oh, we must be wrong. It turned out it was a math weakness [LAUGHTER] that, you know, we later understood that that was an area of, weirdly, of incredible weakness of those early models. But, you know, that was when I realized, OK, the age of cheap intelligence was at its beginning.  LEE: Yeah. So I guess it seems like you had something similar to me in that my first encounters, I actually harbored some skepticism. Is it fair to say you were skeptical before that?  GATES: Well, the idea that we’ve figured out how to encode and access knowledge in this very deep sense without even understanding the nature of the encoding, …  LEE: Right.   GATES: … that is a bit weird.   LEE: Yeah.  GATES: We have an algorithm that creates the computation, but even say, OK, where is the president’s birthday stored in there? Where is this fact stored in there? The fact that even now when we’re playing around, getting a little bit more sense of it, it’s opaque to us what the semantic encoding is, it’s, kind of, amazing to me. I thought the invention of knowledge storage would be an explicit way of encoding knowledge, not an implicit statistical training.  LEE: Yeah, yeah. All right. So, Seb, you know, on this same topic, you know, I got—as we say at Microsoft—I got pulled into the tent. [LAUGHS]  BUBECK: Yes.   LEE: Because this was a very secret project. And then, um, I had the opportunity to select a small number of researchers in MSR [Microsoft Research] to join and start investigating this thing seriously. And the first person I pulled in was you.  BUBECK: Yeah.  LEE: And so what were your first encounters? Because I actually don’t remember what happened then.  BUBECK: Oh, I remember it very well. [LAUGHS] My first encounter with GPT-4 was in a meeting with the two of you, actually. But my kind of first contact, the first moment where I realized that something was happening with generative AI, was before that. And I agree with Bill that I also wasn’t too impressed by GPT-3.  I though that it was kind of, you know, very naturally mimicking the web, sort of parroting what was written there in a nice way. Still in a way which seemed very impressive. But it wasn’t really intelligent in any way. But shortly after GPT-3, there was a model before GPT-4 that really shocked me, and this was the first image generation model, DALL-E 1.  So that was in 2021. And I will forever remember the press release of OpenAI where they had this prompt of an avocado chair and then you had this image of the avocado chair. [LAUGHTER] And what really shocked me is that clearly the model kind of “understood” what is a chair, what is an avocado, and was able to merge those concepts.  So this was really, to me, the first moment where I saw some understanding in those models.   LEE: So this was, just to get the timing right, that was before I pulled you into the tent.  BUBECK: That was before. That was like a year before.  LEE: Right.   BUBECK: And now I will tell you how, you know, we went from that moment to the meeting with the two of you and GPT-4.  So once I saw this kind of understanding, I thought, OK, fine. It understands concept, but it’s still not able to reason. It cannot—as, you know, Bill was saying—it cannot learn from your document. It cannot reason.   So I set out to try to prove that. You know, this is what I was in the business of at the time, trying to prove things in mathematics. So I was trying to prove that basically autoregressive transformers could never reason. So I was trying to prove this. And after a year of work, I had something reasonable to show. And so I had the meeting with the two of you, and I had this example where I wanted to say, there is no way that an LLM is going to be able to do x.  And then as soon as I … I don’t know if you remember, Bill. But as soon as I said that, you said, oh, but wait a second. I had, you know, the OpenAI crew at my house recently, and they showed me a new model. Why don’t we ask this new model this question?   LEE: Yeah. BUBECK: And we did, and it solved it on the spot. And that really, honestly, just changed my life. Like, you know, I had been working for a year trying to say that this was impossible. And just right there, it was shown to be possible.   LEE: [LAUGHS] One of the very first things I got interested in—because I was really thinking a lot about healthcare—was healthcare and medicine.  And I don’t know if the two of you remember, but I ended up doing a lot of tests. I ran through, you know, step one and step two of the US Medical Licensing Exam. Did a whole bunch of other things. I wrote this big report. It was, you know, I can’t remember … a couple hundred pages.   And I needed to share this with someone. I didn’t … there weren’t too many people I could share it with. So I sent, I think, a copy to you, Bill. Sent a copy to you, Seb.   I hardly slept for about a week putting that report together. And, yeah, and I kept working on it. But I was far from alone. I think everyone who was in the tent, so to speak, in those early days was going through something pretty similar. All right. So I think … of course, a lot of what I put in the report also ended up being examples that made it into the book.  But the main purpose of this conversation isn’t to reminisce about [LAUGHS] or indulge in those reminiscences but to talk about what’s happening in healthcare and medicine. And, you know, as I said, we wrote this book. We did it very, very quickly. Seb, you helped. Bill, you know, you provided a review and some endorsements.  But, you know, honestly, we didn’t know what we were talking about because no one had access to this thing. And so we just made a bunch of guesses. So really, the whole thing I wanted to probe with the two of you is, now with two years of experience out in the world, what, you know, what do we think is happening today?  You know, is AI actually having an impact, positive or negative, on healthcare and medicine? And what do we now think is going to happen in the next two years, five years, or 10 years? And so I realize it’s a little bit too abstract to just ask it that way. So let me just try to narrow the discussion and guide us a little bit.   Um, the kind of administrative and clerical work, paperwork, around healthcare—and we made a lot of guesses about that—that appears to be going well, but, you know, Bill, I know we’ve discussed that sometimes that you think there ought to be a lot more going on. Do you have a viewpoint on how AI is actually finding its way into reducing paperwork?  GATES: Well, I’m stunned … I don’t think there should be a patient-doctor meeting where the AI is not sitting in and both transcribing, offering to help with the paperwork, and even making suggestions, although the doctor will be the one, you know, who makes the final decision about the diagnosis and whatever prescription gets done.   It’s so helpful. You know, when that patient goes home and their, you know, son who wants to understand what happened has some questions, that AI should be available to continue that conversation. And the way you can improve that experience and streamline things and, you know, involve the people who advise you. I don’t understand why that’s not more adopted, because there you still have the human in the loop making that final decision.  But even for, like, follow-up calls to make sure the patient did things, to understand if they have concerns and knowing when to escalate back to the doctor, the benefit is incredible. And, you know, that thing is ready for prime time. That paradigm is ready for prime time, in my view.  LEE: Yeah, there are some good products, but it seems like the number one use right now—and we kind of got this from some of the previous guests in previous episodes—is the use of AI just to respond to emails from patients. [LAUGHTER] Does that make sense to you?  BUBECK: Yeah. So maybe I want to second what Bill was saying but maybe take a step back first. You know, two years ago, like, the concept of clinical scribes, which is one of the things that we’re talking about right now, it would have sounded, in fact, it sounded two years ago, borderline dangerous. Because everybody was worried about hallucinations. What happened if you have this AI listening in and then it transcribes, you know, something wrong?  Now, two years later, I think it’s mostly working. And in fact, it is not yet, you know, fully adopted. You’re right. But it is in production. It is used, you know, in many, many places. So this rate of progress is astounding because it wasn’t obvious that we would be able to overcome those obstacles of hallucination. It’s not to say that hallucinations are fully solved. In the case of the closed system, they are.   Now, I think more generally what’s going on in the background is that there is something that we, that certainly I, underestimated, which is this management overhead. So I think the reason why this is not adopted everywhere is really a training and teaching aspect. People need to be taught, like, those systems, how to interact with them.  And one example that I really like, a study that recently appeared where they tried to use ChatGPT for diagnosis and they were comparing doctors without and with ChatGPT (opens in new tab). And the amazing thing … so this was a set of cases where the accuracy of the doctors alone was around 75%. ChatGPT alone was 90%. So that’s already kind of mind blowing. But then the kicker is that doctors with ChatGPT was 80%.   Intelligence alone is not enough. It’s also how it’s presented, how you interact with it. And ChatGPT, it’s an amazing tool. Obviously, I absolutely love it. But it’s not … you don’t want a doctor to have to type in, you know, prompts and use it that way.  It should be, as Bill was saying, kind of running continuously in the background, sending you notifications. And you have to be really careful of the rate at which those notifications are being sent. Because if they are too frequent, then the doctor will learn to ignore them. So you have to … all of those things matter, in fact, at least as much as the level of intelligence of the machine.  LEE: One of the things I think about, Bill, in that scenario that you described, doctors do some thinking about the patient when they write the note. So, you know, I’m always a little uncertain whether it’s actually … you know, you wouldn’t necessarily want to fully automate this, I don’t think. Or at least there needs to be some prompt to the doctor to make sure that the doctor puts some thought into what happened in the encounter with the patient. Does that make sense to you at all?  GATES: At this stage, you know, I’d still put the onus on the doctor to write the conclusions and the summary and not delegate that.  The tradeoffs you make a little bit are somewhat dependent on the situation you’re in. If you’re in Africa, So, yes, the doctor’s still going to have to do a lot of work, but just the quality of letting the patient and the people around them interact and ask questions and have things explained, that alone is such a quality improvement. It’s mind blowing.   LEE: So since you mentioned, you know, Africa—and, of course, this touches on the mission and some of the priorities of the Gates Foundation and this idea of democratization of access to expert medical care—what’s the most interesting stuff going on right now? Are there people and organizations or technologies that are impressing you or that you’re tracking?  GATES: Yeah. So the Gates Foundation has given out a lot of grants to people in Africa doing education, agriculture but more healthcare examples than anything. And the way these things start off, they often start out either being patient-centric in a narrow situation, like, OK, I’m a pregnant woman; talk to me. Or, I have infectious disease symptoms; talk to me. Or they’re connected to a health worker where they’re helping that worker get their job done. And we have lots of pilots out, you know, in both of those cases.   The dream would be eventually to have the thing the patient consults be so broad that it’s like having a doctor available who understands the local things.   LEE: Right.   GATES: We’re not there yet. But over the next two or three years, you know, particularly given the worsening financial constraints against African health systems, where the withdrawal of money has been dramatic, you know, figuring out how to take this—what I sometimes call “free intelligence”—and build a quality health system around that, we will have to be more radical in low-income countries than any rich country is ever going to be.   LEE: Also, there’s maybe a different regulatory environment, so some of those things maybe are easier? Because right now, I think the world hasn’t figured out how to and whether to regulate, let’s say, an AI that might give a medical diagnosis or write a prescription for a medication.  BUBECK: Yeah. I think one issue with this, and it’s also slowing down the deployment of AI in healthcare more generally, is a lack of proper benchmark. Because, you know, you were mentioning the USMLE [United States Medical Licensing Examination], for example. That’s a great test to test human beings and their knowledge of healthcare and medicine. But it’s not a great test to give to an AI.  It’s not asking the right questions. So finding what are the right questions to test whether an AI system is ready to give diagnosis in a constrained setting, that’s a very, very important direction, which to my surprise, is not yet accelerating at the rate that I was hoping for.  LEE: OK, so that gives me an excuse to get more now into the core AI tech because something I’ve discussed with both of you is this issue of what are the right tests. And you both know the very first test I give to any new spin of an LLM is I present a patient, the results—a mythical patient—the results of my physical exam, my mythical physical exam. Maybe some results of some initial labs. And then I present or propose a differential diagnosis. And if you’re not in medicine, a differential diagnosis you can just think of as a prioritized list of the possible diagnoses that fit with all that data. And in that proposed differential, I always intentionally make two mistakes.  I make a textbook technical error in one of the possible elements of the differential diagnosis, and I have an error of omission. And, you know, I just want to know, does the LLM understand what I’m talking about? And all the good ones out there do now. But then I want to know, can it spot the errors? And then most importantly, is it willing to tell me I’m wrong, that I’ve made a mistake?   That last piece seems really hard for AI today. And so let me ask you first, Seb, because at the time of this taping, of course, there was a new spin of GPT-4o last week that became overly sycophantic. In other words, it was actually prone in that test of mine not only to not tell me I’m wrong, but it actually praised me for the creativity of my differential. [LAUGHTER] What’s up with that?  BUBECK: Yeah, I guess it’s a testament to the fact that training those models is still more of an art than a science. So it’s a difficult job. Just to be clear with the audience, we have rolled back that [LAUGHS] version of GPT-4o, so now we don’t have the sycophant version out there.  Yeah, no, it’s a really difficult question. It has to do … as you said, it’s very technical. It has to do with the post-training and how, like, where do you nudge the model? So, you know, there is this very classical by now technique called RLHF [reinforcement learning from human feedback], where you push the model in the direction of a certain reward model. So the reward model is just telling the model, you know, what behavior is good, what behavior is bad.  But this reward model is itself an LLM, and, you know, Bill was saying at the very beginning of the conversation that we don’t really understand how those LLMs deal with concepts like, you know, where is the capital of France located? Things like that. It is the same thing for this reward model. We don’t know why it says that it prefers one output to another, and whether this is correlated with some sycophancy is, you know, something that we discovered basically just now. That if you push too hard in optimization on this reward model, you will get a sycophant model.  So it’s kind of … what I’m trying to say is we became too good at what we were doing, and we ended up, in fact, in a trap of the reward model.  LEE: I mean, you do want … it’s a difficult balance because you do want models to follow your desires and …  BUBECK: It’s a very difficult, very difficult balance.  LEE: So this brings up then the following question for me, which is the extent to which we think we’ll need to have specially trained models for things. So let me start with you, Bill. Do you have a point of view on whether we will need to, you know, quote-unquote take AI models to med school? Have them specially trained? Like, if you were going to deploy something to give medical care in underserved parts of the world, do we need to do something special to create those models?  GATES: We certainly need to teach them the African languages and the unique dialects so that the multimedia interactions are very high quality. We certainly need to teach them the disease prevalence and unique disease patterns like, you know, neglected tropical diseases and malaria. So we need to gather a set of facts that somebody trying to go for a US customer base, you know, wouldn’t necessarily have that in there.  Those two things are actually very straightforward because the additional training time is small. I’d say for the next few years, we’ll also need to do reinforcement learning about the context of being a doctor and how important certain behaviors are. Humans learn over the course of their life to some degree that, I’m in a different context and the way I behave in terms of being willing to criticize or be nice, you know, how important is it? Who’s here? What’s my relationship to them?   Right now, these machines don’t have that broad social experience. And so if you know it’s going to be used for health things, a lot of reinforcement learning of the very best humans in that context would still be valuable. Eventually, the models will, having read all the literature of the world about good doctors, bad doctors, it’ll understand as soon as you say, “I want you to be a doctor diagnosing somebody.” All of the implicit reinforcement that fits that situation, you know, will be there. LEE: Yeah. GATES: And so I hope three years from now, we don’t have to do that reinforcement learning. But today, for any medical context, you would want a lot of data to reinforce tone, willingness to say things when, you know, there might be something significant at stake.  LEE: Yeah. So, you know, something Bill said, kind of, reminds me of another thing that I think we missed, which is, the context also … and the specialization also pertains to different, I guess, what we still call “modes,” although I don’t know if the idea of multimodal is the same as it was two years ago. But, you know, what do you make of all of the hubbub around—in fact, within Microsoft Research, this is a big deal, but I think we’re far from alone—you know, medical images and vision, video, proteins and molecules, cell, you know, cellular data and so on.  BUBECK: Yeah. OK. So there is a lot to say to everything … to the last, you know, couple of minutes. Maybe on the specialization aspect, you know, I think there is, hiding behind this, a really fundamental scientific question of whether eventually we have a singular AGI [artificial general intelligence] that kind of knows everything and you can just put, you know, explain your own context and it will just get it and understand everything.  That’s one vision. I have to say, I don’t particularly believe in this vision. In fact, we humans are not like that at all. I think, hopefully, we are general intelligences, yet we have to specialize a lot. And, you know, I did myself a lot of RL, reinforcement learning, on mathematics. Like, that’s what I did, you know, spent a lot of time doing that. And I didn’t improve on other aspects. You know, in fact, I probably degraded in other aspects. [LAUGHTER] So it’s … I think it’s an important example to have in mind.  LEE: I think I might disagree with you on that, though, because, like, doesn’t a model have to see both good science and bad science in order to be able to gain the ability to discern between the two?  BUBECK: Yeah, no, that absolutely. I think there is value in seeing the generality, in having a very broad base. But then you, kind of, specialize on verticals. And this is where also, you know, open-weights model, which we haven’t talked about yet, are really important because they allow you to provide this broad base to everyone. And then you can specialize on top of it.  LEE: So we have about three hours of stuff to talk about, but our time is actually running low. BUBECK: Yes, yes, yes.   LEE: So I think I want … there’s a more provocative question. It’s almost a silly question, but I need to ask it of the two of you, which is, is there a future, you know, where AI replaces doctors or replaces, you know, medical specialties that we have today? So what does the world look like, say, five years from now?  GATES: Well, it’s important to distinguish healthcare discovery activity from healthcare delivery activity. We focused mostly on delivery. I think it’s very much within the realm of possibility that the AI is not only accelerating healthcare discovery but substituting for a lot of the roles of, you know, I’m an organic chemist, or I run various types of assays. I can see those, which are, you know, testable-output-type jobs but with still very high value, I can see, you know, some replacement in those areas before the doctor.   The doctor, still understanding the human condition and long-term dialogues, you know, they’ve had a lifetime of reinforcement of that, particularly when you get into areas like mental health. So I wouldn’t say in five years, either people will choose to adopt it, but it will be profound that there’ll be this nearly free intelligence that can do follow-up, that can help you, you know, make sure you went through different possibilities.  And so I’d say, yes, we’ll have doctors, but I’d say healthcare will be massively transformed in its quality and in efficiency by AI in that time period.  LEE: Is there a comparison, useful comparison, say, between doctors and, say, programmers, computer programmers, or doctors and, I don’t know, lawyers?  GATES: Programming is another one that has, kind of, a mathematical correctness to it, you know, and so the objective function that you’re trying to reinforce to, as soon as you can understand the state machines, you can have something that’s “checkable”; that’s correct. So I think programming, you know, which is weird to say, that the machine will beat us at most programming tasks before we let it take over roles that have deep empathy, you know, physical presence and social understanding in them.  LEE: Yeah. By the way, you know, I fully expect in five years that AI will produce mathematical proofs that are checkable for validity, easily checkable, because they’ll be written in a proof-checking language like Lean or something but will be so complex that no human mathematician can understand them. I expect that to happen.   I can imagine in some fields, like cellular biology, we could have the same situation in the future because the molecular pathways, the chemistry, biochemistry of human cells or living cells is as complex as any mathematics, and so it seems possible that we may be in a state where in wet lab, we see, Oh yeah, this actually works, but no one can understand why.  BUBECK: Yeah, absolutely. I mean, I think I really agree with Bill’s distinction of the discovery and the delivery, and indeed, the discovery’s when you can check things, and at the end, there is an artifact that you can verify. You know, you can run the protocol in the wet lab and see [if you have] produced what you wanted. So I absolutely agree with that.   And in fact, you know, we don’t have to talk five years from now. I don’t know if you know, but just recently, there was a paper that was published on a scientific discovery using o3- mini (opens in new tab). So this is really amazing. And, you know, just very quickly, just so people know, it was about this statistical physics model, the frustrated Potts model, which has to do with coloring, and basically, the case of three colors, like, more than two colors was open for a long time, and o3 was able to reduce the case of three colors to two colors.   LEE: Yeah.  BUBECK: Which is just, like, astounding. And this is not … this is now. This is happening right now. So this is something that I personally didn’t expect it would happen so quickly, and it’s due to those reasoning models.   Now, on the delivery side, I would add something more to it for the reason why doctors and, in fact, lawyers and coders will remain for a long time, and it’s because we still don’t understand how those models generalize. Like, at the end of the day, we are not able to tell you when they are confronted with a really new, novel situation, whether they will work or not.  Nobody is able to give you that guarantee. And I think until we understand this generalization better, we’re not going to be willing to just let the system in the wild without human supervision.  LEE: But don’t human doctors, human specialists … so, for example, a cardiologist sees a patient in a certain way that a nephrologist …  BUBECK: Yeah. LEE: … or an endocrinologist might not. BUBECK: That’s right. But another cardiologist will understand and, kind of, expect a certain level of generalization from their peer. And this, we just don’t have it with AI models. Now, of course, you’re exactly right. That generalization is also hard for humans. Like, if you have a human trained for one task and you put them into another task, then you don’t … you often don’t know. LEE: OK. You know, the podcast is focused on what’s happened over the last two years. But now, I’d like one provocative prediction about what you think the world of AI and medicine is going to be at some point in the future. You pick your timeframe. I don’t care if it’s two years or 20 years from now, but, you know, what do you think will be different about AI in medicine in that future than today?  BUBECK: Yeah, I think the deployment is going to accelerate soon. Like, we’re really not missing very much. There is this enormous capability overhang. Like, even if progress completely stopped, with current systems, we can do a lot more than what we’re doing right now. So I think this will … this has to be realized, you know, sooner rather than later.  And I think it’s probably dependent on these benchmarks and proper evaluation and tying this with regulation. So these are things that take time in human society and for good reason. But now we already are at two years; you know, give it another two years and it should be really …   LEE: Will AI prescribe your medicines? Write your prescriptions?  BUBECK: I think yes. I think yes.  LEE: OK. Bill?  GATES: Well, I think the next two years, we’ll have massive pilots, and so the amount of use of the AI, still in a copilot-type mode, you know, we should get millions of patient visits, you know, both in general medicine and in the mental health side, as well. And I think that’s going to build up both the data and the confidence to give the AI some additional autonomy. You know, are you going to let it talk to you at night when you’re panicked about your mental health with some ability to escalate? And, you know, I’ve gone so far as to tell politicians with national health systems that if they deploy AI appropriately, that the quality of care, the overload of the doctors, the improvement in the economics will be enough that their voters will be stunned because they just don’t expect this, and, you know, they could be reelected [LAUGHTER] just on this one thing of fixing what is a very overloaded and economically challenged health system in these rich countries.  You know, my personal role is going to be to make sure that in the poorer countries, there isn’t some lag; in fact, in many cases, that we’ll be more aggressive because, you know, we’re comparing to having no access to doctors at all. And, you know, so I think whether it’s India or Africa, there’ll be lessons that are globally valuable because we need medical intelligence. And, you know, thank god AI is going to provide a lot of that.  LEE: Well, on that optimistic note, I think that’s a good way to end. Bill, Seb, really appreciate all of this.   I think the most fundamental prediction we made in the book is that AI would actually find its way into the practice of medicine, and I think that that at least has come true, maybe in different ways than we expected, but it’s come true, and I think it’ll only accelerate from here. So thanks again, both of you.  [TRANSITION MUSIC]  GATES: Yeah. Thanks, you guys.  BUBECK: Thank you, Peter. Thanks, Bill.  LEE: I just always feel such a sense of privilege to have a chance to interact and actually work with people like Bill and Sébastien.    With Bill, I’m always amazed at how practically minded he is. He’s really thinking about the nuts and bolts of what AI might be able to do for people, and his thoughts about underserved parts of the world, the idea that we might actually be able to empower people with access to expert medical knowledge, I think is both inspiring and amazing.   And then, Seb, Sébastien Bubeck, he’s just absolutely a brilliant mind. He has a really firm grip on the deep mathematics of artificial intelligence and brings that to bear in his research and development work. And where that mathematics takes him isn’t just into the nuts and bolts of algorithms but into philosophical questions about the nature of intelligence.   One of the things that Sébastien brought up was the state of evaluation of AI systems. And indeed, he was fairly critical in our conversation. But of course, the world of AI research and development is just moving so fast, and indeed, since we recorded our conversation, OpenAI, in fact, released a new evaluation metric that is directly relevant to medical applications, and that is something called HealthBench. And Microsoft Research also released a new evaluation approach or process called ADeLe.   HealthBench and ADeLe are examples of new approaches to evaluating AI models that are less about testing their knowledge and ability to pass multiple-choice exams and instead are evaluation approaches designed to assess how well AI models are able to complete tasks that actually arise every day in typical healthcare or biomedical research settings. These are examples of really important good work that speak to how well AI models work in the real world of healthcare and biomedical research and how well they can collaborate with human beings in those settings.  You know, I asked Bill and Seb to make some predictions about the future. You know, my own answer, I expect that we’re going to be able to use AI to change how we diagnose patients, change how we decide treatment options.   If you’re a doctor or a nurse and you encounter a patient, you’ll ask questions, do a physical exam, you know, call out for labs just like you do today, but then you’ll be able to engage with AI based on all of that data and just ask, you know, based on all the other people who have gone through the same experience, who have similar data, how were they diagnosed? How were they treated? What were their outcomes? And what does that mean for the patient I have right now? Some people call it the “patients like me” paradigm. And I think that’s going to become real because of AI within our lifetimes. That idea of really grounding the delivery in healthcare and medical practice through data and intelligence, I actually now don’t see any barriers to that future becoming real.  [THEME MUSIC]  I’d like to extend another big thank you to Bill and Sébastien for their time. And to our listeners, as always, it’s a pleasure to have you along for the ride. I hope you’ll join us for our remaining conversations, as well as a second coauthor roundtable with Carey and Zak.   Until next time.   [MUSIC FADES]
    0 Commentarii 0 Distribuiri
  • The art of two Mickeys

    Classic splitscreens, traditional face replacements and new approaches to machine learning-assisted face swapping allowed for twinning shots in ‘Mickey 17’. An excerpt from issue #32 of befores & afters magazine.
    The art of representing two characters on screen at the same time has become known as ‘twinning’. For Mickey 17 visual effects supervisor Dan Glass, the effect of seeing both Mickey 17 and 18 together was one he looked to achieve with a variety of methodologies. “With a technique like that,” he says, “you always want to use a range of tricks, because you don’t want people to figure it out. You want to keep them like, ‘Oh, wait a minute. How did they…?”
    “Going back to the way that Director Bong is so prepared and organized,” adds Glass, “it again makes the world of difference with that kind of work, because he thumbnails every shot. Then, some of them are a bit more fleshed out in storyboards. You can look at it and go, ‘Okay, in this situation, this is what the camera’s doing, this is what the actor’s doing,’ which in itself is quite interesting, because he pre-thinks all of this. You’d think that the actors show up and basically just have to follow the steps like robots. It’s not like that. He gives them an environment to work in, but the shots do end up extraordinarily close to what he thumbnails, and it made it a lot simpler to go through.”

    Those different approaches to twinning ranged from simple splitscreens, to traditional face replacements, and then substantially with a machine learned AI approach, now usually termed ‘face swapping’. What made the twinning work a tougher task than usual, suggests Glass, was the fact that the two Pattinson characters are virtually identical.
    “Normally, when you’re doing some kind of face replacement, you’re comparing it to a memory of the face. But this was right in front of you as two Mickeys looking strikingly similar.”
    Here’s how a typical twinning shot was achieved, as described by Glass. “Because Mickey was mostly dressed the same, with only a slight hair change, we were able to have Robert play both roles and to do them one after another. Sometimes, you have to do these things where hair and makeup or costume has a significant variation, so you’re either waiting a long time, which slows production, or you’re coming back at another time to do the different roles, which always makes the process a lot more complicated to match, but we were able to do that immediately.”

    “Based on the design of the shot,” continues Glass, “I would recommend which of Robert’s parts should be shot first. This was most often determined by which role had more impact on the camera movement. A huge credit goes to Robert for his ability to flip between the roles so effortlessly.”
    In the film, Mickey 17 is more passive and Mickey 18 is more aggressive. Pattinson reflected the distinct characters in his actions, including for a moment in which they fight. This fight, overseen by stunt coordinator Paul Lowe, represented moments of close interaction between the two Mickeys. It was here that a body double was crucial in shooting. The body double was also relied upon for the classic twinning technique of shooting ‘dirty’ over-the- shoulder out of focus shots of the double—ie. 17 looking at 18. However, it was quickly determined that even these would need face replacement work. “Robert’s jawline is so distinct that even those had to be replaced or shot as split screens,” observes Glass.

    When the shot was a moving one, no motion control was employed. “I’ve never been a big advocate for motion control,” states Glass. “To me it’s applicable when you’re doing things like miniatures where you need many matching passes, but I think when performances are involved, it interferes too much. It slows down a production’s speed of movement, but it’s also restrictive. Performance and camera always benefit from more flexibility.”
    “It helped tremendously that Director Bong and DOP Darius Khondji shot quite classically with minimal crane and Steadicam moves,” says Glass. “So, a lot of the moves are pan and dolly. There are some Steadicams in there that we were sometimes able to do splitscreens on. I wasn’t always sure that we could get away with the splitscreen as we shot it, but since we were always shooting the two roles, we had the footage to assess the practicality later. We were always prepared to go down a CG or machine learning route, but where we could use the splitscreen, that was the preference.”
    The Hydralite rig, developed by Volucap. Source:
    Rising Sun Pictureshandled the majority of twinning visual effects, completing them as splitscreen composites, 2D face replacements, and most notably via their machine learning toolset REVIZE, which utilized facial and body capture of Pattinson to train a model of his face and torso to swap for the double’s. A custom capture rig, dubbed the ‘Crazy Rig’ and now officially, The Hydralite, was devised and configured by Volucap to capture multiple angles of Robert on set in each lighting environment in order to produce the best possible reference for the machine learning algorithm. “For me, it was a completely legitimate use of the technique,” attests Glass, in terms of the machine learning approach. “All of the footage that we used to go into that process was captured on our movie for our movie. There’s nothing historic, or going through past libraries of footage, and it was all with Robert’s approval. I think the results were tremendous.”
    “It’s staggering to me as I watch the movie that the performances of each character are so flawlessly consistent throughout the film, because I know how much we were jumping around,” notes Glass. “I did encourage that we rehearse scenes ahead. Let’s say 17 was going to be the first role we captured, I’d have them rehearse it the other way around so that the double knew what he was going to do. Therefore, eyelines, movement, pacing and in instances where we were basically replacing the likeness of his head or even torso, we were still able to use the double’s performance and then map to that.”

    Read the full Mickey 17 issue of befores & afters magazine in PRINT from Amazon or as a DIGITAL EDITION on Patreon. Remember, you can also subscribe to the DIGITAL EDITION as a tier on the Patreon and get a new issue every time one is released.
    The post The art of two Mickeys appeared first on befores & afters.
    #art #two #mickeys
    The art of two Mickeys
    Classic splitscreens, traditional face replacements and new approaches to machine learning-assisted face swapping allowed for twinning shots in ‘Mickey 17’. An excerpt from issue #32 of befores & afters magazine. The art of representing two characters on screen at the same time has become known as ‘twinning’. For Mickey 17 visual effects supervisor Dan Glass, the effect of seeing both Mickey 17 and 18 together was one he looked to achieve with a variety of methodologies. “With a technique like that,” he says, “you always want to use a range of tricks, because you don’t want people to figure it out. You want to keep them like, ‘Oh, wait a minute. How did they…?” “Going back to the way that Director Bong is so prepared and organized,” adds Glass, “it again makes the world of difference with that kind of work, because he thumbnails every shot. Then, some of them are a bit more fleshed out in storyboards. You can look at it and go, ‘Okay, in this situation, this is what the camera’s doing, this is what the actor’s doing,’ which in itself is quite interesting, because he pre-thinks all of this. You’d think that the actors show up and basically just have to follow the steps like robots. It’s not like that. He gives them an environment to work in, but the shots do end up extraordinarily close to what he thumbnails, and it made it a lot simpler to go through.” Those different approaches to twinning ranged from simple splitscreens, to traditional face replacements, and then substantially with a machine learned AI approach, now usually termed ‘face swapping’. What made the twinning work a tougher task than usual, suggests Glass, was the fact that the two Pattinson characters are virtually identical. “Normally, when you’re doing some kind of face replacement, you’re comparing it to a memory of the face. But this was right in front of you as two Mickeys looking strikingly similar.” Here’s how a typical twinning shot was achieved, as described by Glass. “Because Mickey was mostly dressed the same, with only a slight hair change, we were able to have Robert play both roles and to do them one after another. Sometimes, you have to do these things where hair and makeup or costume has a significant variation, so you’re either waiting a long time, which slows production, or you’re coming back at another time to do the different roles, which always makes the process a lot more complicated to match, but we were able to do that immediately.” “Based on the design of the shot,” continues Glass, “I would recommend which of Robert’s parts should be shot first. This was most often determined by which role had more impact on the camera movement. A huge credit goes to Robert for his ability to flip between the roles so effortlessly.” In the film, Mickey 17 is more passive and Mickey 18 is more aggressive. Pattinson reflected the distinct characters in his actions, including for a moment in which they fight. This fight, overseen by stunt coordinator Paul Lowe, represented moments of close interaction between the two Mickeys. It was here that a body double was crucial in shooting. The body double was also relied upon for the classic twinning technique of shooting ‘dirty’ over-the- shoulder out of focus shots of the double—ie. 17 looking at 18. However, it was quickly determined that even these would need face replacement work. “Robert’s jawline is so distinct that even those had to be replaced or shot as split screens,” observes Glass. When the shot was a moving one, no motion control was employed. “I’ve never been a big advocate for motion control,” states Glass. “To me it’s applicable when you’re doing things like miniatures where you need many matching passes, but I think when performances are involved, it interferes too much. It slows down a production’s speed of movement, but it’s also restrictive. Performance and camera always benefit from more flexibility.” “It helped tremendously that Director Bong and DOP Darius Khondji shot quite classically with minimal crane and Steadicam moves,” says Glass. “So, a lot of the moves are pan and dolly. There are some Steadicams in there that we were sometimes able to do splitscreens on. I wasn’t always sure that we could get away with the splitscreen as we shot it, but since we were always shooting the two roles, we had the footage to assess the practicality later. We were always prepared to go down a CG or machine learning route, but where we could use the splitscreen, that was the preference.” The Hydralite rig, developed by Volucap. Source: Rising Sun Pictureshandled the majority of twinning visual effects, completing them as splitscreen composites, 2D face replacements, and most notably via their machine learning toolset REVIZE, which utilized facial and body capture of Pattinson to train a model of his face and torso to swap for the double’s. A custom capture rig, dubbed the ‘Crazy Rig’ and now officially, The Hydralite, was devised and configured by Volucap to capture multiple angles of Robert on set in each lighting environment in order to produce the best possible reference for the machine learning algorithm. “For me, it was a completely legitimate use of the technique,” attests Glass, in terms of the machine learning approach. “All of the footage that we used to go into that process was captured on our movie for our movie. There’s nothing historic, or going through past libraries of footage, and it was all with Robert’s approval. I think the results were tremendous.” “It’s staggering to me as I watch the movie that the performances of each character are so flawlessly consistent throughout the film, because I know how much we were jumping around,” notes Glass. “I did encourage that we rehearse scenes ahead. Let’s say 17 was going to be the first role we captured, I’d have them rehearse it the other way around so that the double knew what he was going to do. Therefore, eyelines, movement, pacing and in instances where we were basically replacing the likeness of his head or even torso, we were still able to use the double’s performance and then map to that.” Read the full Mickey 17 issue of befores & afters magazine in PRINT from Amazon or as a DIGITAL EDITION on Patreon. Remember, you can also subscribe to the DIGITAL EDITION as a tier on the Patreon and get a new issue every time one is released. The post The art of two Mickeys appeared first on befores & afters. #art #two #mickeys
    BEFORESANDAFTERS.COM
    The art of two Mickeys
    Classic splitscreens, traditional face replacements and new approaches to machine learning-assisted face swapping allowed for twinning shots in ‘Mickey 17’. An excerpt from issue #32 of befores & afters magazine. The art of representing two characters on screen at the same time has become known as ‘twinning’. For Mickey 17 visual effects supervisor Dan Glass, the effect of seeing both Mickey 17 and 18 together was one he looked to achieve with a variety of methodologies. “With a technique like that,” he says, “you always want to use a range of tricks, because you don’t want people to figure it out. You want to keep them like, ‘Oh, wait a minute. How did they…?” “Going back to the way that Director Bong is so prepared and organized,” adds Glass, “it again makes the world of difference with that kind of work, because he thumbnails every shot. Then, some of them are a bit more fleshed out in storyboards. You can look at it and go, ‘Okay, in this situation, this is what the camera’s doing, this is what the actor’s doing,’ which in itself is quite interesting, because he pre-thinks all of this. You’d think that the actors show up and basically just have to follow the steps like robots. It’s not like that. He gives them an environment to work in, but the shots do end up extraordinarily close to what he thumbnails, and it made it a lot simpler to go through.” Those different approaches to twinning ranged from simple splitscreens, to traditional face replacements, and then substantially with a machine learned AI approach, now usually termed ‘face swapping’. What made the twinning work a tougher task than usual, suggests Glass, was the fact that the two Pattinson characters are virtually identical. “Normally, when you’re doing some kind of face replacement, you’re comparing it to a memory of the face. But this was right in front of you as two Mickeys looking strikingly similar.” Here’s how a typical twinning shot was achieved, as described by Glass. “Because Mickey was mostly dressed the same, with only a slight hair change, we were able to have Robert play both roles and to do them one after another. Sometimes, you have to do these things where hair and makeup or costume has a significant variation, so you’re either waiting a long time, which slows production, or you’re coming back at another time to do the different roles, which always makes the process a lot more complicated to match, but we were able to do that immediately.” “Based on the design of the shot,” continues Glass, “I would recommend which of Robert’s parts should be shot first. This was most often determined by which role had more impact on the camera movement. A huge credit goes to Robert for his ability to flip between the roles so effortlessly.” In the film, Mickey 17 is more passive and Mickey 18 is more aggressive. Pattinson reflected the distinct characters in his actions, including for a moment in which they fight. This fight, overseen by stunt coordinator Paul Lowe, represented moments of close interaction between the two Mickeys. It was here that a body double was crucial in shooting. The body double was also relied upon for the classic twinning technique of shooting ‘dirty’ over-the- shoulder out of focus shots of the double—ie. 17 looking at 18. However, it was quickly determined that even these would need face replacement work. “Robert’s jawline is so distinct that even those had to be replaced or shot as split screens,” observes Glass. When the shot was a moving one, no motion control was employed. “I’ve never been a big advocate for motion control,” states Glass. “To me it’s applicable when you’re doing things like miniatures where you need many matching passes, but I think when performances are involved, it interferes too much. It slows down a production’s speed of movement, but it’s also restrictive. Performance and camera always benefit from more flexibility.” “It helped tremendously that Director Bong and DOP Darius Khondji shot quite classically with minimal crane and Steadicam moves,” says Glass. “So, a lot of the moves are pan and dolly. There are some Steadicams in there that we were sometimes able to do splitscreens on. I wasn’t always sure that we could get away with the splitscreen as we shot it, but since we were always shooting the two roles, we had the footage to assess the practicality later. We were always prepared to go down a CG or machine learning route, but where we could use the splitscreen, that was the preference.” The Hydralite rig, developed by Volucap. Source: https://volucap.com Rising Sun Pictures (visual effects supervisor Guido Wolter) handled the majority of twinning visual effects, completing them as splitscreen composites, 2D face replacements, and most notably via their machine learning toolset REVIZE, which utilized facial and body capture of Pattinson to train a model of his face and torso to swap for the double’s. A custom capture rig, dubbed the ‘Crazy Rig’ and now officially, The Hydralite, was devised and configured by Volucap to capture multiple angles of Robert on set in each lighting environment in order to produce the best possible reference for the machine learning algorithm. “For me, it was a completely legitimate use of the technique,” attests Glass, in terms of the machine learning approach. “All of the footage that we used to go into that process was captured on our movie for our movie. There’s nothing historic, or going through past libraries of footage, and it was all with Robert’s approval. I think the results were tremendous.” “It’s staggering to me as I watch the movie that the performances of each character are so flawlessly consistent throughout the film, because I know how much we were jumping around,” notes Glass. “I did encourage that we rehearse scenes ahead. Let’s say 17 was going to be the first role we captured, I’d have them rehearse it the other way around so that the double knew what he was going to do. Therefore, eyelines, movement, pacing and in instances where we were basically replacing the likeness of his head or even torso, we were still able to use the double’s performance and then map to that.” Read the full Mickey 17 issue of befores & afters magazine in PRINT from Amazon or as a DIGITAL EDITION on Patreon. Remember, you can also subscribe to the DIGITAL EDITION as a tier on the Patreon and get a new issue every time one is released. The post The art of two Mickeys appeared first on befores & afters.
    0 Commentarii 0 Distribuiri
  • '28 Years Later' used 20 iPhones in tandem for some wild shots

    It's no secret that 28 Years Later used iPhones to shoot parts of the film. Now its director, Danny Boyle, has discussed the use of iPhones for the film in more detail with IGN. The first film in the franchise, 28 Days Later, was shot on digital video, giving it a homemade feel. Boyle explained that he and writer Alex Garland got the idea from the fact that home video cameras were common at the time, and people would've shot videos with them if an apocalypse had indeed happened. Those cameras, of course, have since been replaced by smartphones.
    The movies used three special rigs for the iPhone sequences: One for eight cameras that one person can carry, another with 10 and another with 20. "I never say this, but there is an incredible shot in the second halfwhere we use the 20-rig camera, and you'll know it when you see it," Boyle told IGN. He described the 20-iPhone rig as "basically a poor man’s bullet time," which is a visual effect that uses multiple cameras to freeze or slow down time. Think the scene in The Matrix, wherein Neo dodged bullets in super slow motion. 
    Doyle said that the 20-camera rig can be attached to cranes or dollies and give you 180 degrees of vision of an action. In editing, you can choose from any of the footage each iPhone takes to, say, move between perspectives or jump forward and backward. For 28 Years Later, the team used the rig for violent scenes to emphasize their effect. "For a moment the audience is inside the scene, the action, rather than classically observing a picture," Doyle explained.
    In addition to the iPhones, the filmmakers also used drones, cameras attached to actors and even farm animals to achieve an immersive feel for its 2.76:1 widescreen aspect ratio. They decided on the aspect ratio to create a sense of unease, since you'd have to keep scanning the screen to see potential threats coming from the sides. 
    Sony
    This article originally appeared on Engadget at
    #years #later039 #used #iphones #tandem
    '28 Years Later' used 20 iPhones in tandem for some wild shots
    It's no secret that 28 Years Later used iPhones to shoot parts of the film. Now its director, Danny Boyle, has discussed the use of iPhones for the film in more detail with IGN. The first film in the franchise, 28 Days Later, was shot on digital video, giving it a homemade feel. Boyle explained that he and writer Alex Garland got the idea from the fact that home video cameras were common at the time, and people would've shot videos with them if an apocalypse had indeed happened. Those cameras, of course, have since been replaced by smartphones. The movies used three special rigs for the iPhone sequences: One for eight cameras that one person can carry, another with 10 and another with 20. "I never say this, but there is an incredible shot in the second halfwhere we use the 20-rig camera, and you'll know it when you see it," Boyle told IGN. He described the 20-iPhone rig as "basically a poor man’s bullet time," which is a visual effect that uses multiple cameras to freeze or slow down time. Think the scene in The Matrix, wherein Neo dodged bullets in super slow motion.  Doyle said that the 20-camera rig can be attached to cranes or dollies and give you 180 degrees of vision of an action. In editing, you can choose from any of the footage each iPhone takes to, say, move between perspectives or jump forward and backward. For 28 Years Later, the team used the rig for violent scenes to emphasize their effect. "For a moment the audience is inside the scene, the action, rather than classically observing a picture," Doyle explained. In addition to the iPhones, the filmmakers also used drones, cameras attached to actors and even farm animals to achieve an immersive feel for its 2.76:1 widescreen aspect ratio. They decided on the aspect ratio to create a sense of unease, since you'd have to keep scanning the screen to see potential threats coming from the sides.  Sony This article originally appeared on Engadget at #years #later039 #used #iphones #tandem
    WWW.ENGADGET.COM
    '28 Years Later' used 20 iPhones in tandem for some wild shots
    It's no secret that 28 Years Later used iPhones to shoot parts of the film. Now its director, Danny Boyle, has discussed the use of iPhones for the film in more detail with IGN. The first film in the franchise, 28 Days Later, was shot on digital video, giving it a homemade feel. Boyle explained that he and writer Alex Garland got the idea from the fact that home video cameras were common at the time, and people would've shot videos with them if an apocalypse had indeed happened. Those cameras, of course, have since been replaced by smartphones. The movies used three special rigs for the iPhone sequences: One for eight cameras that one person can carry, another with 10 and another with 20. "I never say this, but there is an incredible shot in the second half [of the film] where we use the 20-rig camera, and you'll know it when you see it," Boyle told IGN. He described the 20-iPhone rig as "basically a poor man’s bullet time," which is a visual effect that uses multiple cameras to freeze or slow down time. Think the scene in The Matrix, wherein Neo dodged bullets in super slow motion.  Doyle said that the 20-camera rig can be attached to cranes or dollies and give you 180 degrees of vision of an action. In editing, you can choose from any of the footage each iPhone takes to, say, move between perspectives or jump forward and backward. For 28 Years Later, the team used the rig for violent scenes to emphasize their effect. "For a moment the audience is inside the scene, the action, rather than classically observing a picture," Doyle explained. In addition to the iPhones, the filmmakers also used drones, cameras attached to actors and even farm animals to achieve an immersive feel for its 2.76:1 widescreen aspect ratio. They decided on the aspect ratio to create a sense of unease, since you'd have to keep scanning the screen to see potential threats coming from the sides.  Sony This article originally appeared on Engadget at https://www.engadget.com/entertainment/28-years-later-used-20-iphones-in-tandem-for-some-wild-shots-130043338.html?src=rss
    0 Commentarii 0 Distribuiri
  • Outlets 8, Conghua by E Plus Design: Chromatic Urbanism and Ecological Renewal

    Outlets 8, Conghua | © Wu Siming
    In the landscape of contemporary Chinese urbanism, few typologies encapsulate the contradictions of late-capitalist development more vividly than the pseudo-European commercial complex. These replicated enclaves, constructed en masse in the early 2000s, were once marketed as symbols of international sophistication. Over time, however, many were abandoned, becoming architectural vestiges of speculative urbanism. Outlets 8 in Conghua, Guangzhou, is one such project that has undergone a radical architectural reinterpretation. Originally completed in 2018 but long dormant, it has been reimagined by E Plus Design in collaboration with URBANUS/LXD Studio. Through a precise, light-touch intervention, the project avoids wholesale demolition and reprograms space through color, rhythm, and landscape strategy.

    Outlets 8, Conghua Technical Information

    Architects1-14: E Plus Design
    Central Plaza Design: URBANUS / LXD Studio
    Location: Conghua District, Guangzhou, China
    Gross Area: 80,882 m2 | 870,000 Sq. Ft.
    Project Years: 2022 – 2023
    Photographs: © Wu Siming

    This approach is like a contemporary remix of classical music. The four blocks correspond to four movements. Without extensive demolition or altering the European-style architectural rhythm, we reinterpreted the emotional tones, chords, and cadenzas. Through a blend of color and modern gestures, the outdated and disproportionate ‘faux-antique’ complex has been reorchestrated into a contemporary architectural symphony.
    – Li Fu, Chief Architect at E Plus Design

    Outlets 8, Conghua Photographs

    Aerial View | © Wu Siming

    © Wu Siming

    © Wu Siming

    © Wu Siming

    © Wu Siming

    © Wu Siming

    © Wu Siming

    © Chen Liang Liu Shan

    © Chen Liang Liu Shan

    © Chen Liang Liu Shan
    Outlets 8 Context and Typological Challenge
    Outlets 8 was initially conceived as a 110,000-square-meter faux-European outlet village. Despite its scale and investment, it struggled to resonate with local cultural dynamics and remained idle. The typology itself, rooted in nostalgic mimicry, was already facing obsolescence. The challenge, then, was not only architectural but also conceptual: how to resuscitate a typology that had become both spatially and culturally inert.
    The design team chose a strategy of minimal physical intervention coupled with maximal perceptual impact. Rather than demolish or drastically reconstruct, they aimed to re-signify the existing structures. This approach reflects a growing trend in urban renewal across China, where sustainability, cost-efficiency, and cultural specificity take precedence over spectacle.
    Spatial Transformation Through Chromatic Reprogramming

    After | © Wu Siming

    Before | Original Facade, © E+

    At the intervention’s core is using color as a spatial and psychological agent. The ornament-heavy facades were stripped of their polychromatic excess and repainted in low-saturation hues. This chromatic cleansing revealed the formal rhythms of the architecture beneath. By doing so, the design avoids mimicry and opts for abstraction, reintroducing clarity to the site’s visual language.
    The design framework is structured as a musical metaphor, with each of the four blocks conceived as a separate movement in a visual symphony. The street-facing facades, now unified through a golden “variation,” establish a new urban frontage that is both legible and symbolically rich. A ribbon-like golden band traces across the main elevations, creating continuity and contrast between old and new volumes.
    In contrast, the sports block adopts a cooler, blue-toned palette, offering a different spatial and functional rhythm. New architectural insertions are rendered in transparent materials, signaling temporal and programmatic distinctions. At the center, the elliptical plaza becomes a spatial crescendo, defined by a sculptural intervention inspired by Roman aqueducts. This feature functions as a landmark and a temporal break, juxtaposing historical references with performative landscape elements.
    Rewriting Landscape as Urban Ecology

    After | © Wu Siming

    Before | Original Facade, © E+

    Water, derived from the nearby Liuxi River, serves as the thematic and material backbone of the landscape design. Its integration is not symbolic but functional. Water flows through constructed channels, interactive fountains, and sculptural cascades that encourage observation and participation. These elements create a multisensory environment that enhances the spatial experience while reinforcing ecological awareness.
    The planting strategy emphasizes native species capable of withstanding Guangzhou’s subtropical climate. The design maximizes greenery wherever regulatory conditions allow, particularly along the main entrance, central corridors, and arcaded walkways. The result is a layered landscape that balances visual density with ecological resilience.
    Integrating landscape and architecture as a singular design operation, the project shifts away from ornamental greening toward environmental synthesis. This approach foregrounds interaction and immersion, aligning with broader shifts in landscape architecture toward performative and participatory ecologies.
    Programmatic Rebirth and Urban Implications

    After | © Wu Siming

    Before | Original Facade, © E+

    Beyond formal and material considerations, the project redefines the programmatic potential of large-scale retail environments. Positioned as a “micro-vacation” destination, Outlets 8 is a hybrid typology. It combines retail, leisure, and outdoor experience within a cohesive spatial narrative. This reprogramming responds to changing patterns of consumption and leisure in Chinese cities, particularly among younger demographics seeking experiential value over transactional efficiency.
    Statistical metrics underscore the project’s social impact. In its first nine days, the outlet attracted over half a million visitors and became a trending location across multiple digital platforms. While not the focus of architectural critique, these figures reflect a successful alignment between spatial renewal and public resonance.
    More importantly, the project offers a replicable model for dealing with the vast inventory of misaligned commercial developments across China. The intervention avoids nostalgia and cynicism by foregrounding perceptual clarity, ecological integration, and cultural recontextualization. Instead, it offers a clear path forward for reimagining the built remnants of a prior urban paradigm.
    Outlets 8, Conghua Plans

    Elevations | © E Plus Design

    Floor Plan | © E Plus Design

    Floor Plan | © E Plus Design

    Floor Plan | © E Plus Design

    Floor Plan | © E Plus Design

    Sections | © E Plus Design
    Outlets 8, Conghua Image Gallery

    About E Plus Design
    E Plus Design is a multidisciplinary architecture studio based in Shenzhen, China, known for its innovative approaches to urban renewal, adaptive reuse, and large-scale public space transformations. The firm emphasizes minimal intervention strategies, spatial clarity, and contextual sensitivity, often working at the intersection of architecture, landscape, and urban design to create integrated environments that are both socially responsive and experientially rich.
    Credits and Additional Notes

    Chief Design Consultant: Liu Xiaodu
    Master Plan, Architecture, and Landscape Schemes: E Plus Design
    Lead Architects: Li Fu, Coco Zhou
    Project Managers: Guo Sibo, Huang Haifeng
    Architectural Design Team: Wang Junli, Zhang Yan, Cai Yidie, Zhu Meng, Lin Zhaomei, Li Geng, Stephane Anil Mamode, Liu Shan, Zhou Yubo
    Central Plaza Design: URBANUS / LXD Studio
    Architect of Central Plaza: Liu Xiaodu
    Project Manager: Li An’hong
    Facade Design: Song Baolin, Li Minggang
    Lighting Design: Fang Yuhui
    Lighting Consultant: Han Du Associates
    Client: Guangzhou Outlets 8 Commercial Management Co., Ltd.
    Client Design Management Team: Yin Mingyue, Zhao Xiong
    Landscape Area: 29,100 m²
    Chief Landscape Architect: Gao Yan
    Project Manager: Zhang Yufeng
    Landscape Design Team: Yu Xiaolei, Li Zhaozhan, Liu Chenghua
    Landscape Construction Drawings: E Plus Design
    Project Manager: Wang Bin
    Design Team: Wang Bin. Huang Jinxiong. Li GenStructural Design Team: Wang Kaiming, Yang Helin, Wu Xingwei, Zhuang Dengfa
    Electrical Design Team: Sun Wei, Yang Ying
    Interior Design Concept Design: Shenzhen Juanshi Design Co., Ltd.
    Chief Interior Designer: Feng Feifan
    Project Manager: Liu Hongwei
    Design Team: Niu Jingxian, Shi Meitao
    Construction Drawings: Shenzhen Shiye Design Co., Ltd.
    Project Manager: Shen Kaizhen
    Design Team: Yao Yijian, Yang Hao, Liu Chen
    Wayfinding Design Studio: Hexi Brand Design Co., Ltd.
    Curtain Wall Design Firm: Positive Attitude Group
    #outlets #conghua #plus #design #chromatic
    Outlets 8, Conghua by E Plus Design: Chromatic Urbanism and Ecological Renewal
    Outlets 8, Conghua | © Wu Siming In the landscape of contemporary Chinese urbanism, few typologies encapsulate the contradictions of late-capitalist development more vividly than the pseudo-European commercial complex. These replicated enclaves, constructed en masse in the early 2000s, were once marketed as symbols of international sophistication. Over time, however, many were abandoned, becoming architectural vestiges of speculative urbanism. Outlets 8 in Conghua, Guangzhou, is one such project that has undergone a radical architectural reinterpretation. Originally completed in 2018 but long dormant, it has been reimagined by E Plus Design in collaboration with URBANUS/LXD Studio. Through a precise, light-touch intervention, the project avoids wholesale demolition and reprograms space through color, rhythm, and landscape strategy. Outlets 8, Conghua Technical Information Architects1-14: E Plus Design Central Plaza Design: URBANUS / LXD Studio Location: Conghua District, Guangzhou, China Gross Area: 80,882 m2 | 870,000 Sq. Ft. Project Years: 2022 – 2023 Photographs: © Wu Siming This approach is like a contemporary remix of classical music. The four blocks correspond to four movements. Without extensive demolition or altering the European-style architectural rhythm, we reinterpreted the emotional tones, chords, and cadenzas. Through a blend of color and modern gestures, the outdated and disproportionate ‘faux-antique’ complex has been reorchestrated into a contemporary architectural symphony. – Li Fu, Chief Architect at E Plus Design Outlets 8, Conghua Photographs Aerial View | © Wu Siming © Wu Siming © Wu Siming © Wu Siming © Wu Siming © Wu Siming © Wu Siming © Chen Liang Liu Shan © Chen Liang Liu Shan © Chen Liang Liu Shan Outlets 8 Context and Typological Challenge Outlets 8 was initially conceived as a 110,000-square-meter faux-European outlet village. Despite its scale and investment, it struggled to resonate with local cultural dynamics and remained idle. The typology itself, rooted in nostalgic mimicry, was already facing obsolescence. The challenge, then, was not only architectural but also conceptual: how to resuscitate a typology that had become both spatially and culturally inert. The design team chose a strategy of minimal physical intervention coupled with maximal perceptual impact. Rather than demolish or drastically reconstruct, they aimed to re-signify the existing structures. This approach reflects a growing trend in urban renewal across China, where sustainability, cost-efficiency, and cultural specificity take precedence over spectacle. Spatial Transformation Through Chromatic Reprogramming After | © Wu Siming Before | Original Facade, © E+ At the intervention’s core is using color as a spatial and psychological agent. The ornament-heavy facades were stripped of their polychromatic excess and repainted in low-saturation hues. This chromatic cleansing revealed the formal rhythms of the architecture beneath. By doing so, the design avoids mimicry and opts for abstraction, reintroducing clarity to the site’s visual language. The design framework is structured as a musical metaphor, with each of the four blocks conceived as a separate movement in a visual symphony. The street-facing facades, now unified through a golden “variation,” establish a new urban frontage that is both legible and symbolically rich. A ribbon-like golden band traces across the main elevations, creating continuity and contrast between old and new volumes. In contrast, the sports block adopts a cooler, blue-toned palette, offering a different spatial and functional rhythm. New architectural insertions are rendered in transparent materials, signaling temporal and programmatic distinctions. At the center, the elliptical plaza becomes a spatial crescendo, defined by a sculptural intervention inspired by Roman aqueducts. This feature functions as a landmark and a temporal break, juxtaposing historical references with performative landscape elements. Rewriting Landscape as Urban Ecology After | © Wu Siming Before | Original Facade, © E+ Water, derived from the nearby Liuxi River, serves as the thematic and material backbone of the landscape design. Its integration is not symbolic but functional. Water flows through constructed channels, interactive fountains, and sculptural cascades that encourage observation and participation. These elements create a multisensory environment that enhances the spatial experience while reinforcing ecological awareness. The planting strategy emphasizes native species capable of withstanding Guangzhou’s subtropical climate. The design maximizes greenery wherever regulatory conditions allow, particularly along the main entrance, central corridors, and arcaded walkways. The result is a layered landscape that balances visual density with ecological resilience. Integrating landscape and architecture as a singular design operation, the project shifts away from ornamental greening toward environmental synthesis. This approach foregrounds interaction and immersion, aligning with broader shifts in landscape architecture toward performative and participatory ecologies. Programmatic Rebirth and Urban Implications After | © Wu Siming Before | Original Facade, © E+ Beyond formal and material considerations, the project redefines the programmatic potential of large-scale retail environments. Positioned as a “micro-vacation” destination, Outlets 8 is a hybrid typology. It combines retail, leisure, and outdoor experience within a cohesive spatial narrative. This reprogramming responds to changing patterns of consumption and leisure in Chinese cities, particularly among younger demographics seeking experiential value over transactional efficiency. Statistical metrics underscore the project’s social impact. In its first nine days, the outlet attracted over half a million visitors and became a trending location across multiple digital platforms. While not the focus of architectural critique, these figures reflect a successful alignment between spatial renewal and public resonance. More importantly, the project offers a replicable model for dealing with the vast inventory of misaligned commercial developments across China. The intervention avoids nostalgia and cynicism by foregrounding perceptual clarity, ecological integration, and cultural recontextualization. Instead, it offers a clear path forward for reimagining the built remnants of a prior urban paradigm. Outlets 8, Conghua Plans Elevations | © E Plus Design Floor Plan | © E Plus Design Floor Plan | © E Plus Design Floor Plan | © E Plus Design Floor Plan | © E Plus Design Sections | © E Plus Design Outlets 8, Conghua Image Gallery About E Plus Design E Plus Design is a multidisciplinary architecture studio based in Shenzhen, China, known for its innovative approaches to urban renewal, adaptive reuse, and large-scale public space transformations. The firm emphasizes minimal intervention strategies, spatial clarity, and contextual sensitivity, often working at the intersection of architecture, landscape, and urban design to create integrated environments that are both socially responsive and experientially rich. Credits and Additional Notes Chief Design Consultant: Liu Xiaodu Master Plan, Architecture, and Landscape Schemes: E Plus Design Lead Architects: Li Fu, Coco Zhou Project Managers: Guo Sibo, Huang Haifeng Architectural Design Team: Wang Junli, Zhang Yan, Cai Yidie, Zhu Meng, Lin Zhaomei, Li Geng, Stephane Anil Mamode, Liu Shan, Zhou Yubo Central Plaza Design: URBANUS / LXD Studio Architect of Central Plaza: Liu Xiaodu Project Manager: Li An’hong Facade Design: Song Baolin, Li Minggang Lighting Design: Fang Yuhui Lighting Consultant: Han Du Associates Client: Guangzhou Outlets 8 Commercial Management Co., Ltd. Client Design Management Team: Yin Mingyue, Zhao Xiong Landscape Area: 29,100 m² Chief Landscape Architect: Gao Yan Project Manager: Zhang Yufeng Landscape Design Team: Yu Xiaolei, Li Zhaozhan, Liu Chenghua Landscape Construction Drawings: E Plus Design Project Manager: Wang Bin Design Team: Wang Bin. Huang Jinxiong. Li GenStructural Design Team: Wang Kaiming, Yang Helin, Wu Xingwei, Zhuang Dengfa Electrical Design Team: Sun Wei, Yang Ying Interior Design Concept Design: Shenzhen Juanshi Design Co., Ltd. Chief Interior Designer: Feng Feifan Project Manager: Liu Hongwei Design Team: Niu Jingxian, Shi Meitao Construction Drawings: Shenzhen Shiye Design Co., Ltd. Project Manager: Shen Kaizhen Design Team: Yao Yijian, Yang Hao, Liu Chen Wayfinding Design Studio: Hexi Brand Design Co., Ltd. Curtain Wall Design Firm: Positive Attitude Group #outlets #conghua #plus #design #chromatic
    ARCHEYES.COM
    Outlets 8, Conghua by E Plus Design: Chromatic Urbanism and Ecological Renewal
    Outlets 8, Conghua | © Wu Siming In the landscape of contemporary Chinese urbanism, few typologies encapsulate the contradictions of late-capitalist development more vividly than the pseudo-European commercial complex. These replicated enclaves, constructed en masse in the early 2000s, were once marketed as symbols of international sophistication. Over time, however, many were abandoned, becoming architectural vestiges of speculative urbanism. Outlets 8 in Conghua, Guangzhou, is one such project that has undergone a radical architectural reinterpretation. Originally completed in 2018 but long dormant, it has been reimagined by E Plus Design in collaboration with URBANUS/LXD Studio. Through a precise, light-touch intervention, the project avoids wholesale demolition and reprograms space through color, rhythm, and landscape strategy. Outlets 8, Conghua Technical Information Architects1-14: E Plus Design Central Plaza Design: URBANUS / LXD Studio Location: Conghua District, Guangzhou, China Gross Area: 80,882 m2 | 870,000 Sq. Ft. Project Years: 2022 – 2023 Photographs: © Wu Siming This approach is like a contemporary remix of classical music. The four blocks correspond to four movements. Without extensive demolition or altering the European-style architectural rhythm, we reinterpreted the emotional tones, chords, and cadenzas. Through a blend of color and modern gestures, the outdated and disproportionate ‘faux-antique’ complex has been reorchestrated into a contemporary architectural symphony. – Li Fu, Chief Architect at E Plus Design Outlets 8, Conghua Photographs Aerial View | © Wu Siming © Wu Siming © Wu Siming © Wu Siming © Wu Siming © Wu Siming © Wu Siming © Chen Liang Liu Shan © Chen Liang Liu Shan © Chen Liang Liu Shan Outlets 8 Context and Typological Challenge Outlets 8 was initially conceived as a 110,000-square-meter faux-European outlet village. Despite its scale and investment, it struggled to resonate with local cultural dynamics and remained idle. The typology itself, rooted in nostalgic mimicry, was already facing obsolescence. The challenge, then, was not only architectural but also conceptual: how to resuscitate a typology that had become both spatially and culturally inert. The design team chose a strategy of minimal physical intervention coupled with maximal perceptual impact. Rather than demolish or drastically reconstruct, they aimed to re-signify the existing structures. This approach reflects a growing trend in urban renewal across China, where sustainability, cost-efficiency, and cultural specificity take precedence over spectacle. Spatial Transformation Through Chromatic Reprogramming After | © Wu Siming Before | Original Facade, © E+ At the intervention’s core is using color as a spatial and psychological agent. The ornament-heavy facades were stripped of their polychromatic excess and repainted in low-saturation hues. This chromatic cleansing revealed the formal rhythms of the architecture beneath. By doing so, the design avoids mimicry and opts for abstraction, reintroducing clarity to the site’s visual language. The design framework is structured as a musical metaphor, with each of the four blocks conceived as a separate movement in a visual symphony. The street-facing facades, now unified through a golden “variation,” establish a new urban frontage that is both legible and symbolically rich. A ribbon-like golden band traces across the main elevations, creating continuity and contrast between old and new volumes. In contrast, the sports block adopts a cooler, blue-toned palette, offering a different spatial and functional rhythm. New architectural insertions are rendered in transparent materials, signaling temporal and programmatic distinctions. At the center, the elliptical plaza becomes a spatial crescendo, defined by a sculptural intervention inspired by Roman aqueducts. This feature functions as a landmark and a temporal break, juxtaposing historical references with performative landscape elements. Rewriting Landscape as Urban Ecology After | © Wu Siming Before | Original Facade, © E+ Water, derived from the nearby Liuxi River, serves as the thematic and material backbone of the landscape design. Its integration is not symbolic but functional. Water flows through constructed channels, interactive fountains, and sculptural cascades that encourage observation and participation. These elements create a multisensory environment that enhances the spatial experience while reinforcing ecological awareness. The planting strategy emphasizes native species capable of withstanding Guangzhou’s subtropical climate. The design maximizes greenery wherever regulatory conditions allow, particularly along the main entrance, central corridors, and arcaded walkways. The result is a layered landscape that balances visual density with ecological resilience. Integrating landscape and architecture as a singular design operation, the project shifts away from ornamental greening toward environmental synthesis. This approach foregrounds interaction and immersion, aligning with broader shifts in landscape architecture toward performative and participatory ecologies. Programmatic Rebirth and Urban Implications After | © Wu Siming Before | Original Facade, © E+ Beyond formal and material considerations, the project redefines the programmatic potential of large-scale retail environments. Positioned as a “micro-vacation” destination, Outlets 8 is a hybrid typology. It combines retail, leisure, and outdoor experience within a cohesive spatial narrative. This reprogramming responds to changing patterns of consumption and leisure in Chinese cities, particularly among younger demographics seeking experiential value over transactional efficiency. Statistical metrics underscore the project’s social impact. In its first nine days, the outlet attracted over half a million visitors and became a trending location across multiple digital platforms. While not the focus of architectural critique, these figures reflect a successful alignment between spatial renewal and public resonance. More importantly, the project offers a replicable model for dealing with the vast inventory of misaligned commercial developments across China. The intervention avoids nostalgia and cynicism by foregrounding perceptual clarity, ecological integration, and cultural recontextualization. Instead, it offers a clear path forward for reimagining the built remnants of a prior urban paradigm. Outlets 8, Conghua Plans Elevations | © E Plus Design Floor Plan | © E Plus Design Floor Plan | © E Plus Design Floor Plan | © E Plus Design Floor Plan | © E Plus Design Sections | © E Plus Design Outlets 8, Conghua Image Gallery About E Plus Design E Plus Design is a multidisciplinary architecture studio based in Shenzhen, China, known for its innovative approaches to urban renewal, adaptive reuse, and large-scale public space transformations. The firm emphasizes minimal intervention strategies, spatial clarity, and contextual sensitivity, often working at the intersection of architecture, landscape, and urban design to create integrated environments that are both socially responsive and experientially rich. Credits and Additional Notes Chief Design Consultant: Liu Xiaodu Master Plan, Architecture, and Landscape Schemes: E Plus Design Lead Architects: Li Fu, Coco Zhou Project Managers (Architecture): Guo Sibo, Huang Haifeng Architectural Design Team: Wang Junli, Zhang Yan, Cai Yidie, Zhu Meng, Lin Zhaomei, Li Geng, Stephane Anil Mamode, Liu Shan, Zhou Yubo Central Plaza Design: URBANUS / LXD Studio Architect of Central Plaza: Liu Xiaodu Project Manager: Li An’hong Facade Design: Song Baolin, Li Minggang Lighting Design (Concept): Fang Yuhui Lighting Consultant: Han Du Associates Client: Guangzhou Outlets 8 Commercial Management Co., Ltd. Client Design Management Team: Yin Mingyue, Zhao Xiong Landscape Area: 29,100 m² Chief Landscape Architect: Gao Yan Project Manager (Landscape): Zhang Yufeng Landscape Design Team: Yu Xiaolei, Li Zhaozhan, Liu Chenghua Landscape Construction Drawings: E Plus Design Project Manager: Wang Bin Design Team: Wang Bin (Landscape Architecture). Huang Jinxiong (Greening Design). Li Gen (Water & Electricity Design) Structural Design Team: Wang Kaiming, Yang Helin, Wu Xingwei, Zhuang Dengfa Electrical Design Team: Sun Wei, Yang Ying Interior Design Concept Design: Shenzhen Juanshi Design Co., Ltd. Chief Interior Designer: Feng Feifan Project Manager: Liu Hongwei Design Team: Niu Jingxian, Shi Meitao Construction Drawings: Shenzhen Shiye Design Co., Ltd. Project Manager: Shen Kaizhen Design Team: Yao Yijian, Yang Hao, Liu Chen Wayfinding Design Studio: Hexi Brand Design Co., Ltd. Curtain Wall Design Firm: Positive Attitude Group (PAG)
    0 Commentarii 0 Distribuiri
  • Gironda Residence by Giovanni Mecozzi: The Renovation of Casa Guaccimanni in Ravenna

    Gironda Residence | © Simone Bossi
    Located just steps from Piazza del Popolo in Ravenna, the Renaissance-era Casa Guaccimanni holds centuries of architectural and historical weight. Constructed in the fifteenth century for the Venetian podestà Nicolò Giustinian, the building evolved through noble ownership and later became home to Vittorio and Alessandro Guaccimanni, sons of Risorgimento figure Luigi Guaccimanni. Architecturally, the structure is characterized by a tripartite plan with a central corridor flanked by large rooms, an interior courtyard with a double loggia, and decorative elements spanning Renaissance to Neoclassical periods. Once concealed beneath plaster, its frescoed veranda and exposed wooden ceilings speak to a layered history of intervention, concealment, and rediscovery.

    Gironda Residence in Casa Guaccimanni Technical Information

    Architects1-13: Giovanni Mecozzi Architetti
    Location: Casa Guaccimanni, Via Armando Diaz, Ravenna, Italy
    Client: Emanuela Docimo
    Project Years: 2022 – 2024
    Original Structure: 15th Century
    Photographs: © Andrea Sestito, © Simone Bossi, © Omar Sartor

    The new and the old never touch, but gently brush against each other, maintaining a distance capable of generating tension.
    – Giovanni Mecozzi

    Gironda Residence in Casa Guaccimanni Photographs

    © Omar Sartor

    © Andrea Sestito

    © Andrea Sestito

    © Andrea Sestito

    © Omar Sartor

    © Simone Bossi

    © Simone Bossi

    © Simone Bossi

    © Omar Sartor

    © Omar Sartor

    © Omar Sartor

    © Andrea Sestito

    © Omar Sartor
    Design Intent: Reversibility and Temporal Tension
    The recent architectural project by Giovanni Mecozzi Architetti centers on the noble floor of the palazzo, reinterpreted as a contemporary residence named Gironda. Rather than imposing a new visual regime onto the historic shell, the intervention operates with restraint, foregrounding the building’s original character while establishing new spatial and material conditions.
    At the core of the project lies a design philosophy rooted in reversibility. Mecozzi’s intervention resists permanence. The furnishings and spatial devices introduced into the historic rooms are self-supporting and detached from the structure. No new element makes physical contact with the floors, ceilings, or walls, preserving the integrity of the original surfaces. This strategy avoids irreversible alterations and allows the architecture to remain temporally flexible.
    Architect Giovanni Mecozzi articulates this approach succinctly: “The new and the old never touch, but gently brush against each other, maintaining a distance capable of generating tension.” This spatial tension is not decorative but conceptual, prompting occupants to consider the relationship between historical continuity and contemporary transformation. The design does not attempt to erase time but rather exposes its layers through careful juxtaposition.
    The project draws conceptual and chromatic inspiration from Ravenna’s early Christian and Byzantine mosaics. Rather than replicate ornamental motifs, Mecozzi extracts abstract qualities such as color, luminosity, and surface texture, integrating them as subtle spatial references throughout the residence.
    Gironda Residence Material Strategy
    Access to the residence is organized through a longitudinal hallway that bisects the plan, connecting a balcony on the north façade with a loggia overlooking the garden to the south. This corridor becomes a spine for circulation and orientation, punctuated by entries into five main rooms: the kitchen, veranda, and three independent suites.
    Each suite functions as a self-contained spatial environment. The original large rooms have been reimagined with integrated volumes housing diverse domestic functions: bathrooms, saunas, walk-in closets, reading nooks, and home cinemas. These new programmatic layers are embedded within freestanding furniture structures, which operate more as inhabitable objects than architectural partitions.
    Color becomes an operative tool for spatial differentiation. The three principal suites, the Gold Room, the Blue Room, and the Green Room, are introduced chromatically through thresholds that face the main corridor. This prelude of color sets the tone for each room’s unique interior experience. Within, glossy glass tiles, gilded surfaces, and a reduced palette of materials establish a scenographic yet restrained environment.
    The flooring, a Venetian terrazzo installed during earlier restoration work in the 2000s, has been retained. Its beveled borders and rounded corners respond to the proportions of each room, reinforcing a visual continuity that binds the new interventions with the inherited context. In contrast to the historical envelope, the furniture and spatial devices employ a language of monochromatic forms and minimal detailing, occasionally verging on neoplastic abstraction. This tension between old ornament and new abstraction is one of the project’s defining features.
    Furnishings curated by Atelier Biagetti, known for their theatrical and ironic sensibility, further enrich the atmosphere. These pieces do not mimic the historical setting but create moments of visual friction and playful ambiguity, enhancing the multi-temporal character of the interiors.
    Architectural Significance and Cultural Dialogue
    The Gironda residence exemplifies a growing discourse in contemporary architecture around adaptive reuse that neither mimics nor erases the past. Rather than treating heritage as a constraint or an aesthetic to be curated, Mecozzi engages it as an active agent in spatial transformation. The project is a case study in reversible architecture, where temporality is embedded in the design, not just its historical references.
    This intervention prompts broader questions about the role of preservation in contemporary practice. Can architectural interventions occupy historic contexts without becoming parasitic or nostalgic? Mecozzi’s project suggests that they can adopt a posture of critical distance and conceptual clarity.
    Gironda does not attempt to restore Casa Guaccimanni to a previous state or impose a singular vision of modernity. Instead, it crafts a dialogue between past and present, structured through spatial strategies, material choices, and chromatic cues. In doing so, it opens a new chapter in the building’s ongoing life, one that is fully contemporary yet deeply rooted in architectural memory.
    Gironda Residence in Casa Guaccimanni Plans

    Floor Plan | © Giovanni Mecozzi Architetti

    Golden Room Layout | © Giovanni Mecozzi Architetti

    Door Detail | © Giovanni Mecozzi Architetti
    Gironda Residence in Casa Guaccimanni Image Gallery

    About Giovanni Mecozzi
    Giovanni Mecozzi is an Italian architect based in Ravenna, Italy, and the founder of Giovanni Mecozzi Architetti, a multidisciplinary studio specializing in architecture, interior design, and landscape projects. After graduating from the University of Ferrara with an architecture degree, Mecozzi gained international experience working in Spain, including collaborating with Mendaro Arquitectos in Madrid. Upon returning to Italy, he co-founded GMA, focusing on projects emphasizing the relationship between architecture, the client, and the context, with a particular interest in renovating and transforming historical buildings. 
    Credits and Additional Notes

    Design Team: Giovanni Mecozzi, Cecilia Verdini, Filippo Minghetti
    Construction: EdilcostruzioniElectrical Systems: Elektra ServiceMechanical and Hydraulic Systems: Nuova OLP
    Structural Alterations: Not applicableCustom Furniture: Idea LegnoCurtains and Fabrics: Selezione Arredamenti, Ravenna
    Lighting: ViabizzunoResin Coatings and Flooring: Kerakoll
    Rugs and Carpeting: Centro Moquette, Rimini
    Bathroom Furnishings: Salaroli, Ravenna
    Furniture, Artwork, and Design Objects Selected by: Atelier BiagettiFurniture Designers: Alberto Biagetti and Laura Baldassarri
    #gironda #residence #giovanni #mecozzi #renovation
    Gironda Residence by Giovanni Mecozzi: The Renovation of Casa Guaccimanni in Ravenna
    Gironda Residence | © Simone Bossi Located just steps from Piazza del Popolo in Ravenna, the Renaissance-era Casa Guaccimanni holds centuries of architectural and historical weight. Constructed in the fifteenth century for the Venetian podestà Nicolò Giustinian, the building evolved through noble ownership and later became home to Vittorio and Alessandro Guaccimanni, sons of Risorgimento figure Luigi Guaccimanni. Architecturally, the structure is characterized by a tripartite plan with a central corridor flanked by large rooms, an interior courtyard with a double loggia, and decorative elements spanning Renaissance to Neoclassical periods. Once concealed beneath plaster, its frescoed veranda and exposed wooden ceilings speak to a layered history of intervention, concealment, and rediscovery. Gironda Residence in Casa Guaccimanni Technical Information Architects1-13: Giovanni Mecozzi Architetti Location: Casa Guaccimanni, Via Armando Diaz, Ravenna, Italy Client: Emanuela Docimo Project Years: 2022 – 2024 Original Structure: 15th Century Photographs: © Andrea Sestito, © Simone Bossi, © Omar Sartor The new and the old never touch, but gently brush against each other, maintaining a distance capable of generating tension. – Giovanni Mecozzi Gironda Residence in Casa Guaccimanni Photographs © Omar Sartor © Andrea Sestito © Andrea Sestito © Andrea Sestito © Omar Sartor © Simone Bossi © Simone Bossi © Simone Bossi © Omar Sartor © Omar Sartor © Omar Sartor © Andrea Sestito © Omar Sartor Design Intent: Reversibility and Temporal Tension The recent architectural project by Giovanni Mecozzi Architetti centers on the noble floor of the palazzo, reinterpreted as a contemporary residence named Gironda. Rather than imposing a new visual regime onto the historic shell, the intervention operates with restraint, foregrounding the building’s original character while establishing new spatial and material conditions. At the core of the project lies a design philosophy rooted in reversibility. Mecozzi’s intervention resists permanence. The furnishings and spatial devices introduced into the historic rooms are self-supporting and detached from the structure. No new element makes physical contact with the floors, ceilings, or walls, preserving the integrity of the original surfaces. This strategy avoids irreversible alterations and allows the architecture to remain temporally flexible. Architect Giovanni Mecozzi articulates this approach succinctly: “The new and the old never touch, but gently brush against each other, maintaining a distance capable of generating tension.” This spatial tension is not decorative but conceptual, prompting occupants to consider the relationship between historical continuity and contemporary transformation. The design does not attempt to erase time but rather exposes its layers through careful juxtaposition. The project draws conceptual and chromatic inspiration from Ravenna’s early Christian and Byzantine mosaics. Rather than replicate ornamental motifs, Mecozzi extracts abstract qualities such as color, luminosity, and surface texture, integrating them as subtle spatial references throughout the residence. Gironda Residence Material Strategy Access to the residence is organized through a longitudinal hallway that bisects the plan, connecting a balcony on the north façade with a loggia overlooking the garden to the south. This corridor becomes a spine for circulation and orientation, punctuated by entries into five main rooms: the kitchen, veranda, and three independent suites. Each suite functions as a self-contained spatial environment. The original large rooms have been reimagined with integrated volumes housing diverse domestic functions: bathrooms, saunas, walk-in closets, reading nooks, and home cinemas. These new programmatic layers are embedded within freestanding furniture structures, which operate more as inhabitable objects than architectural partitions. Color becomes an operative tool for spatial differentiation. The three principal suites, the Gold Room, the Blue Room, and the Green Room, are introduced chromatically through thresholds that face the main corridor. This prelude of color sets the tone for each room’s unique interior experience. Within, glossy glass tiles, gilded surfaces, and a reduced palette of materials establish a scenographic yet restrained environment. The flooring, a Venetian terrazzo installed during earlier restoration work in the 2000s, has been retained. Its beveled borders and rounded corners respond to the proportions of each room, reinforcing a visual continuity that binds the new interventions with the inherited context. In contrast to the historical envelope, the furniture and spatial devices employ a language of monochromatic forms and minimal detailing, occasionally verging on neoplastic abstraction. This tension between old ornament and new abstraction is one of the project’s defining features. Furnishings curated by Atelier Biagetti, known for their theatrical and ironic sensibility, further enrich the atmosphere. These pieces do not mimic the historical setting but create moments of visual friction and playful ambiguity, enhancing the multi-temporal character of the interiors. Architectural Significance and Cultural Dialogue The Gironda residence exemplifies a growing discourse in contemporary architecture around adaptive reuse that neither mimics nor erases the past. Rather than treating heritage as a constraint or an aesthetic to be curated, Mecozzi engages it as an active agent in spatial transformation. The project is a case study in reversible architecture, where temporality is embedded in the design, not just its historical references. This intervention prompts broader questions about the role of preservation in contemporary practice. Can architectural interventions occupy historic contexts without becoming parasitic or nostalgic? Mecozzi’s project suggests that they can adopt a posture of critical distance and conceptual clarity. Gironda does not attempt to restore Casa Guaccimanni to a previous state or impose a singular vision of modernity. Instead, it crafts a dialogue between past and present, structured through spatial strategies, material choices, and chromatic cues. In doing so, it opens a new chapter in the building’s ongoing life, one that is fully contemporary yet deeply rooted in architectural memory. Gironda Residence in Casa Guaccimanni Plans Floor Plan | © Giovanni Mecozzi Architetti Golden Room Layout | © Giovanni Mecozzi Architetti Door Detail | © Giovanni Mecozzi Architetti Gironda Residence in Casa Guaccimanni Image Gallery About Giovanni Mecozzi Giovanni Mecozzi is an Italian architect based in Ravenna, Italy, and the founder of Giovanni Mecozzi Architetti, a multidisciplinary studio specializing in architecture, interior design, and landscape projects. After graduating from the University of Ferrara with an architecture degree, Mecozzi gained international experience working in Spain, including collaborating with Mendaro Arquitectos in Madrid. Upon returning to Italy, he co-founded GMA, focusing on projects emphasizing the relationship between architecture, the client, and the context, with a particular interest in renovating and transforming historical buildings.  Credits and Additional Notes Design Team: Giovanni Mecozzi, Cecilia Verdini, Filippo Minghetti Construction: EdilcostruzioniElectrical Systems: Elektra ServiceMechanical and Hydraulic Systems: Nuova OLP Structural Alterations: Not applicableCustom Furniture: Idea LegnoCurtains and Fabrics: Selezione Arredamenti, Ravenna Lighting: ViabizzunoResin Coatings and Flooring: Kerakoll Rugs and Carpeting: Centro Moquette, Rimini Bathroom Furnishings: Salaroli, Ravenna Furniture, Artwork, and Design Objects Selected by: Atelier BiagettiFurniture Designers: Alberto Biagetti and Laura Baldassarri #gironda #residence #giovanni #mecozzi #renovation
    ARCHEYES.COM
    Gironda Residence by Giovanni Mecozzi: The Renovation of Casa Guaccimanni in Ravenna
    Gironda Residence | © Simone Bossi Located just steps from Piazza del Popolo in Ravenna, the Renaissance-era Casa Guaccimanni holds centuries of architectural and historical weight. Constructed in the fifteenth century for the Venetian podestà Nicolò Giustinian, the building evolved through noble ownership and later became home to Vittorio and Alessandro Guaccimanni, sons of Risorgimento figure Luigi Guaccimanni. Architecturally, the structure is characterized by a tripartite plan with a central corridor flanked by large rooms, an interior courtyard with a double loggia, and decorative elements spanning Renaissance to Neoclassical periods. Once concealed beneath plaster, its frescoed veranda and exposed wooden ceilings speak to a layered history of intervention, concealment, and rediscovery. Gironda Residence in Casa Guaccimanni Technical Information Architects1-13: Giovanni Mecozzi Architetti Location: Casa Guaccimanni, Via Armando Diaz, Ravenna, Italy Client: Emanuela Docimo Project Years: 2022 – 2024 Original Structure: 15th Century Photographs: © Andrea Sestito, © Simone Bossi, © Omar Sartor The new and the old never touch, but gently brush against each other, maintaining a distance capable of generating tension. – Giovanni Mecozzi Gironda Residence in Casa Guaccimanni Photographs © Omar Sartor © Andrea Sestito © Andrea Sestito © Andrea Sestito © Omar Sartor © Simone Bossi © Simone Bossi © Simone Bossi © Omar Sartor © Omar Sartor © Omar Sartor © Andrea Sestito © Omar Sartor Design Intent: Reversibility and Temporal Tension The recent architectural project by Giovanni Mecozzi Architetti centers on the noble floor of the palazzo, reinterpreted as a contemporary residence named Gironda. Rather than imposing a new visual regime onto the historic shell, the intervention operates with restraint, foregrounding the building’s original character while establishing new spatial and material conditions. At the core of the project lies a design philosophy rooted in reversibility. Mecozzi’s intervention resists permanence. The furnishings and spatial devices introduced into the historic rooms are self-supporting and detached from the structure. No new element makes physical contact with the floors, ceilings, or walls, preserving the integrity of the original surfaces. This strategy avoids irreversible alterations and allows the architecture to remain temporally flexible. Architect Giovanni Mecozzi articulates this approach succinctly: “The new and the old never touch, but gently brush against each other, maintaining a distance capable of generating tension.” This spatial tension is not decorative but conceptual, prompting occupants to consider the relationship between historical continuity and contemporary transformation. The design does not attempt to erase time but rather exposes its layers through careful juxtaposition. The project draws conceptual and chromatic inspiration from Ravenna’s early Christian and Byzantine mosaics. Rather than replicate ornamental motifs, Mecozzi extracts abstract qualities such as color, luminosity, and surface texture, integrating them as subtle spatial references throughout the residence. Gironda Residence Material Strategy Access to the residence is organized through a longitudinal hallway that bisects the plan, connecting a balcony on the north façade with a loggia overlooking the garden to the south. This corridor becomes a spine for circulation and orientation, punctuated by entries into five main rooms: the kitchen, veranda, and three independent suites. Each suite functions as a self-contained spatial environment. The original large rooms have been reimagined with integrated volumes housing diverse domestic functions: bathrooms, saunas, walk-in closets, reading nooks, and home cinemas. These new programmatic layers are embedded within freestanding furniture structures, which operate more as inhabitable objects than architectural partitions. Color becomes an operative tool for spatial differentiation. The three principal suites, the Gold Room, the Blue Room, and the Green Room, are introduced chromatically through thresholds that face the main corridor. This prelude of color sets the tone for each room’s unique interior experience. Within, glossy glass tiles, gilded surfaces, and a reduced palette of materials establish a scenographic yet restrained environment. The flooring, a Venetian terrazzo installed during earlier restoration work in the 2000s, has been retained. Its beveled borders and rounded corners respond to the proportions of each room, reinforcing a visual continuity that binds the new interventions with the inherited context. In contrast to the historical envelope, the furniture and spatial devices employ a language of monochromatic forms and minimal detailing, occasionally verging on neoplastic abstraction. This tension between old ornament and new abstraction is one of the project’s defining features. Furnishings curated by Atelier Biagetti, known for their theatrical and ironic sensibility, further enrich the atmosphere. These pieces do not mimic the historical setting but create moments of visual friction and playful ambiguity, enhancing the multi-temporal character of the interiors. Architectural Significance and Cultural Dialogue The Gironda residence exemplifies a growing discourse in contemporary architecture around adaptive reuse that neither mimics nor erases the past. Rather than treating heritage as a constraint or an aesthetic to be curated, Mecozzi engages it as an active agent in spatial transformation. The project is a case study in reversible architecture, where temporality is embedded in the design, not just its historical references. This intervention prompts broader questions about the role of preservation in contemporary practice. Can architectural interventions occupy historic contexts without becoming parasitic or nostalgic? Mecozzi’s project suggests that they can adopt a posture of critical distance and conceptual clarity. Gironda does not attempt to restore Casa Guaccimanni to a previous state or impose a singular vision of modernity. Instead, it crafts a dialogue between past and present, structured through spatial strategies, material choices, and chromatic cues. In doing so, it opens a new chapter in the building’s ongoing life, one that is fully contemporary yet deeply rooted in architectural memory. Gironda Residence in Casa Guaccimanni Plans Floor Plan | © Giovanni Mecozzi Architetti Golden Room Layout | © Giovanni Mecozzi Architetti Door Detail | © Giovanni Mecozzi Architetti Gironda Residence in Casa Guaccimanni Image Gallery About Giovanni Mecozzi Giovanni Mecozzi is an Italian architect based in Ravenna, Italy, and the founder of Giovanni Mecozzi Architetti (GMA), a multidisciplinary studio specializing in architecture, interior design, and landscape projects. After graduating from the University of Ferrara with an architecture degree, Mecozzi gained international experience working in Spain, including collaborating with Mendaro Arquitectos in Madrid. Upon returning to Italy, he co-founded GMA, focusing on projects emphasizing the relationship between architecture, the client, and the context, with a particular interest in renovating and transforming historical buildings.  Credits and Additional Notes Design Team: Giovanni Mecozzi, Cecilia Verdini, Filippo Minghetti Construction: Edilcostruzioni (Leoni Andrea) Electrical Systems: Elektra Service (Andrea Baiardi) Mechanical and Hydraulic Systems: Nuova OLP Structural Alterations: Not applicable (intervention is fully reversible) Custom Furniture: Idea Legno (Paolo Berdondini) Curtains and Fabrics: Selezione Arredamenti, Ravenna Lighting: Viabizzuno (via Tutto Luce, Cesena) Resin Coatings and Flooring: Kerakoll Rugs and Carpeting: Centro Moquette, Rimini Bathroom Furnishings: Salaroli, Ravenna Furniture, Artwork, and Design Objects Selected by: Atelier Biagetti (Milan) Furniture Designers: Alberto Biagetti and Laura Baldassarri
    0 Commentarii 0 Distribuiri