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]

html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd" Japanese architect Shigeru Ban has created the Blue Ocean Dome (BOD) for the Osaka-Kansai Expo 2025, addressing the urgent issue of marine plastic pollution and raising crucial awareness about it.Named Blue Ocean Dome, the pavilion stands out with its innovative design, comprising three distinct dome types: Dome A, Dome B, and Dome C. Each dome is specifically crafted to host captivating installations and dynamic exhibitions, promising an unforgettable experience for all visitors throughout the event. Image © Taiki FukaoThe project was commissioned by the Zero Emissions Research and Initiatives (ZERI), a global network of creative minds, seeking solutions to the ever increasing problems of the world.Rather than outright rejecting plastic, the pavilion inspires deep reflection on how we use and manage materials, highlighting our critical responsibility to make sustainable choices for the future.The BOD merges traditional and modern materials—like bamboo, paper, and carbon fiber reinforced plastic (CFRP)—to unlock new and innovative architectural possibilities.Dome A, serving as the striking entrance, is expertly crafted from laminated bamboo. This innovative design not only showcases the beauty of bamboo but also tackles the pressing issue of abandoned bamboo groves in Japan, which pose a risk to land stability due to their shallow root systems.Utilizing raw bamboo for structural purposes is often difficult; however, through advanced processing, it is transformed into thin, laminated boards that boast strength even greater than that of conventional wood. These boards have been skillfully fashioned into a remarkable 19-meter dome, drawing inspiration from traditional Japanese bamboo hats. This project brilliantly turns an environmental challenge into a sustainable architectural solution, highlighting the potential of bamboo as a valuable resource.Dome B stands as the central and largest structure of its kind, boasting a remarkable diameter of 42 meters. It is primarily constructed from Carbon Fiber Reinforced Polymer (CFRP), a cutting-edge material revered for its extraordinary strength-to-weight ratio—four times stronger than steel yet only one-fifth the weight. While CFRP is predominantly seen in industries such as aerospace and automotive due to its high cost, its application in architecture is pioneering.In this project, the choice of CFRP was not just advantageous; it was essential. The primary goal was to minimize the foundation weight on the reclaimed land of the Expo site, making sustainability a top priority. To mitigate the environmental consequences of deep foundation piles, the structure had to be lighter than the soil excavated for its foundation. CFRP not only met this stringent requirement but also ensured the dome's structural integrity, showcasing a perfect marriage of innovation and environmental responsibility.Dome C, with its impressive 19-meter diameter, is crafted entirely from paper tubes that are 100% recyclable after use. Its innovative design features a three-dimensional truss structure, connected by elegant wooden spheres, evoking the beauty of molecular structures.To champion sustainability and minimize waste following the six-month Expo, the entire BOD pavilion has been meticulously designed for effortless disassembly and relocation. It is anchored by a robust steel foundation system and boasts a modular design that allows it to be conveniently packed into standard shipping containers. After the Expo concludes, this remarkable pavilion will be transported to the Maldives, where it will be transformed into a stunning resort facility, breathing new life into its design and purpose.Recently, Shigeru Ban's Paper Log House was revealed at Philip Johnson's Glass House Venue. In addition, Ban installed his Paper Partition Shelters (PPS) for the victims of the Turkey-Syria earthquake in Mersin and Hatay provinces of Turkey.All images © Hiroyuki Hirai unless otherwise stated.> via Shigeru Ban Architects 

This edition of the Venice Biennale includes 65 national pavilions, 11 collateral events, and over 750 participants in the international exhibition curated by Italian architect and engineer Carlo Ratti. Entitled Intelligens: Natural Artificial Collective, its stated aim is to make Venice a ‘living laboratory’. But Ratti’s exhibition in the Arsenale has been hit by mixed reviews. The AJ’s Rob Wilson described it as ‘a bit of a confusing mess’, while other media outlets have called the robot-heavy exhibit of future-facing building-focused solutions to the climate crisis a ‘tech-bro fever dream’ and a ‘mind-boggling rollercoaster’ to mention a few. It is a distinct shift away from the biennale of two years ago twhen Ghanaian-Scottish architect Lesley Lokko curated the main exhibitions, including 89 participants – of which more than half were from Africa or the African diaspora – in a convincing reset of the architectural conversation.Advertisement This year’s National Pavilions and collateral exhibits, by contrast, have tackled the largest themes in architecture and the world right now in a less constrained way than the main exhibitions. The exhibits are radical and work as a useful gauge for understanding what’s important in each country: decarbonisation, climate resilience, the reconstruction of Gaza, and an issue more prevalent in politics closer to home: gender wars. What's not to miss in the Giardini? British Pavilion (photography: Chris Lane) UK Pavilion The British Pavilion this year, which won a special mention from the Venetian jury, is housing a show by a British-Kenyan collab titled GBR – Geology of Britannic Repair. In it, the curators explore the links between colonialism, the built environment and geological extraction. Focusing on the Rift Valley, which runs from east Africa to the Middle East, including Palestine, the exhibition was curated by the Nairobi-based studio cave_bureau, UK-based curator, writer and Farrell Centre director Owen Hopkins and Queen Mary University professor Kathryn Yusoff. The pavilion’s façade is cloaked by a beaded veil of agricultural waste briquettes and clay and glass beads, produced in Kenya and India, echoing both Maasai practices and beads once made on Venice’s Murano, as currency for the exchange of metals, minerals and slaves. The pavilion’s six gallery spaces include multisensory installations such as the Earth Compass, a series of celestial maps connecting London and Nairobi; the Rift Room, tracing one of humans’ earliest migration routes; and the Shimoni Slave Cave, featuring a large-scale bronze cast of a valley cave historically used as a holding pen for enslaved people.Advertisement The show also includes Objects of Repair, a project by design-led research group Palestine Regeneration Team (PART), looking at how salvaged materials could help rebuild war-torn Gaza, the only exhibit anywhere in the Biennale that tackled the reconstruction of Gaza face-on – doing so impressively, both politically and sensitively. Read more here. Danish Pavilion (photography: Hampus Berndtson) Demark Pavilion A firm favourite by most this year, the Danish exhibition Build of Site, curated by Søren Pihlmann of Pihlmann Architects, transforms the pavilion, which requires renovation anyway, into both a renovation site and archive of materials. Clever, simple and very methodical, the building is being both renewed while at the same time showcasing innovative methods to reuse surplus materials uncovered during the construction process – as an alternative to using new resources to build a temporary exhibition. The renovation of the 1950s Peter Koch-designed section of the pavilion began in December 2024 and will be completed following the biennale, having been suspended for its duration. On display are archetypal elements including podiums, ramps, benches and tables – all constructed from the surplus materials unearthed during the renovation, such as wood, limestone, concrete, stone, sand, silt and clay. Belgian Pavilion (photography: Michiel De Cleene) Belgium Pavilion If you need a relaxing break from the intensity of the biennale, then the oldest national pavilion in the Giardini is the one for you. Belgium’s Building Biospheres: A New Alliance between Nature and Architecture brings ‘plant intelligence’ to the fore. Commissioned by the Flanders Architecture Institute and curated by landscape architect Bas Smets and neurobiologist Stefano Mancuso, the exhibit investigates how the natural ‘intelligence’ of plants can be used to produce an indoor climate – elevating the role of landscape design and calling for it to no longer serve as a backdrop for architecture. Inside, more than 200 plants occupy the central area beneath the skylight, becoming the pavilion’s centrepiece, with the rear space visualising ‘real-time’ data on the prototype’s climate control performance. Spanish Pavilion (photography: Luca Capuano) Spain Pavilion One for the pure architecture lovers out there, models (32!), installations, photographs and timber structures fill the Spanish Pavilion in abundance. Neatly curated by architects Roi Salgueiro Barrio and Manuel Bouzas Barcala, Internalities shows a series of existing and research projects that have contributed to decarbonising construction in Spain. The outcome? An extensive collection of work exploring the use of very local and very specific regenerative and low-carbon construction and materials – including stone, wood and soil. The joy of this pavilion comes from the 16 beautiful timber frames constructed from wood from communal forests in Galicia. Polish Pavilion (photography: Luca Capuano) Poland Pavilion Poland’s pavilion was like Marmite this year. Some loved its playful approach while others found it silly. Lares and Penates, taking its name from ancient Roman deities of protection, has been curated by Aleksandra Kędziorek and looks at what it means and takes to have a sense of security in architecture. Speaking to many different anxieties, it refers to the unspoken assumption of treating architecture as a safe haven against the elements, catastrophes and wars – showcasing and elevating the mundane solutions and signage derived from building, fire and health regulations. The highlight? An ornate niche decorated with tiles and stones just for … a fire extinguisher. Dutch Pavilion (photography: Cristiano Corte) Netherlands Pavilion Punchy and straight to the point, SIDELINED: A Space to Rethink Togetherness takes sports as a lens for looking at how spatial design can both reveal and disrupt the often-exclusionary dynamics of everyday environments. Within the pavilion, the exhibit looks beyond the large-scale arena of the stadium and gymnasium to investigate the more localised and intimate context of the sports bar, as well as three alternative sports – a site of both social production and identity formation – as a metaphor for uniting diverse communities. The pavilion-turned-sports bar, designed by Koos Breen and Jeannette Slütter and inspired by Asger Jorn’s three-sided sports field, is a space for fluidity and experimentation where binary oppositions, social hierarchies and cultural values are contested and reshaped – complete with jerseys and football scarfs (currently a must-have fashion item) worn by players in the alternative Anonymous Allyship aligning the walls. Read Derin Fadina’s review for the AJ here. Performance inside the Nordic Countries Pavilion (photography: Venla Helenius) Nordic Countries Pavilion Probably the most impactful national pavilion this year (and with the best tote bag by far), the Nordic Countries have presented an installation with performance work. Curated by Kaisa Karvinen, Industry Muscle: Five Scores for Architecture continues Finnish artist Teo Ala-Ruona’s work on trans embodiment and ecology by considering the trans body as a lens through which to examine modern architecture and the built environment. The three-day exhibition opening featured a two-hour performance each day with Ala-Ruona and his troupe crawling, climbing and writhing around the space, creating a bodily dialogue with the installations and pavilion building itself, which was designed by celebrated Modernist architect Sverre Fehn. The American pavilion next door, loudly (country music!) turns its back on what’s going on in its own country by just celebrating the apathetical porch, making the Nordic Countries seem even more relevant in this crucial time. Read Derin Fadina’s review for the AJ here. German Pavilion (photography: Luca Capuano) Germany Pavilion An exhibit certainly grabbing the issue of climate change by its neck is the German contribution, Stresstest. Curated by Nicola Borgmann, Elisabeth Endres, Gabriele G Kiefer and Daniele Santucci, the pavilion has turned climate change into a literal physical and psychological experience for visitors by creating contrasting ‘stress’ and ‘de-stress’ rooms. In the dark stress room, a large metal sculpture creates a cramped and hot space using heating mats hung from the ceiling and powered by PVs. Opposite is a calmer space demonstrating strategies that could be used to reduce the heat of cities, and between the two spaces is a film focusing on the impacts of cities becoming hotter. If this doesn’t highlight the urgency of the situation, I’m not sure what will. Best bits of the Arsenale outside the main exhibitions Bahrain Pavilion (photography: Andrea Avezzù) Bahrain Pavilion Overall winner of this year’s Golden Lion for best national participation, Bahrain’s pavilion in the historic Artiglierie of the Arsenale is a proposal for living and working through heat conditions. Heatwave, curated by architect Andrea Faraguna, reimagines public space design by exploring passive cooling strategies rooted in the Arab country’s climate, as well as cultural context. A geothermal well and solar chimney are connected through a thermo-hygrometric axis that links underground conditions with the air outside. The inhabitable space that hosts visitors is thus compressed and defined by its earth-covered floor and suspended ceiling, and is surrounded by memorable sandbags, highlighting its scalability for particularly hot construction sites in the Gulf where a huge amount of construction is taking place. In the Arsenale’s exhibition space, where excavation wasn’t feasible, this system has been adapted into mechanical ventilation, bringing in air from the canal side and channelling it through ductwork to create a microclimate. Slovenian Pavilion (photography: Andrea Avezzù) Slovenia Pavilion The AJ’s Rob Wilson’s top pavilion tip this year provides an enjoyable take on the theme of the main exhibition, highlighting how the tacit knowledge and on-site techniques and skills of construction workers and craftspeople are still the key constituent in architectural production despite all the heat and light about robotics, prefabrication, artificial intelligence and 3D printing. Master Builders, curated by Ana Kosi and Ognen Arsov and organised by the Museum of Architecture and Design (MAO) in Ljubljana, presents a series of ‘totems’ –accumulative sculpture-like structures that are formed of conglomerations of differently worked materials, finishes and building elements. These are stacked up into crazy tower forms, which showcase various on-site construction skills and techniques, their construction documented in accompanying films. Uzbekistan Pavilion (photography: Luca Capuano) Uzbekistan Pavilion Uzbekistan’s contribution explores the Soviet era solar furnace and Modernist legacy. Architecture studio GRACE, led by curators Ekaterina Golovatyuk and Giacomo Cantoni have curated A Matter of Radiance. The focus is the Sun Institute of Material Science – originally known as the Sun Heliocomplex – an incredible large-scale scientific structure built in 1987 on a natural, seismic-free foundation near Tashkent and one of only two that study material behaviour under extreme temperatures. The exhibition examines the solar oven’s site’s historical and contemporary significance while reflecting on its scientific legacy and influence moving beyond just national borders. Applied Arts Pavilion (photography: Andrea Avezzù) V&A Applied Arts Pavilion Diller Scofidio + Renfro (DS+R) is having a moment. The US-based practice, in collaboration with V&A chief curator Brendan Cormier, has curated On Storage, which aptly explores global storage architectures in a pavilion that strongly links to the V&A’s recent opening of Storehouse, its new (and free) collections archive in east London. Featured is a six-channel (and screen) film entitled Boxed: The Mild Boredom of Order, directed by the practice itself and following a toothbrush, as a metaphor for an everyday consumer product, on its journey through different forms of storage across the globe – from warehouse to distribution centre to baggage handlers down to the compact space of a suitcase. Also on display are large-format photographs of V&A East Storehouse, DS+R’s original architectural model and sketchbook and behind-the-scenes photography of Storehouse at work, taken by emerging east London-based photographers. Canal Café (photography: Marco Zorzanello) Canal café Golden Lion for the best participation in the actual exhibition went to Canal Café, an intervention designed by V&A East Storehouse’s architect DS+R with Natural Systems Utilities, SODAI, Aaron Betsky and Davide Oldani. Serving up canal-water espresso, the installation is a demonstration of how Venice itself can be a laboratory to understand how to live on the water in a time of water scarcity. The structure, located on the edge of the Arsenale’s building complex, draws water from its lagoon before filtering it onsite via a hybrid of natural and artificial methods, including a mini wetland with grasses. The project was recognised for its persistence, having started almost 20 years ago, just showing how water scarcity, contamination and flooding are still major concerns both globally and, more locally, in the tourist-heavy city of Venice. And what else? Holy See Pavilion (photography: Andrea Avezzù) The Holy See Much like the Danish Pavilion, the Pavilion of the Holy See is also taking on an approach of renewal this year. Over the next six months, Opera Aperta will breathe new life into the Santa Maria Ausiliatrice Complex in the Castello district of Venice. Founded as a hospice for pilgrims in 1171, the building later became the oldest hospital and was converted into school in the 18th century. In 2001, the City of Venice allocated it for cultural use and for the next four years it will be managed by the Dicastery for Culture and Education of the Holy See to oversee its restoration. Curated by architect, curator and researcher Marina Otero Verzier and artistic director of Fondaco Italia, Giovanna Zabotti, the complex has been turned into a constant ‘living laboratory’ of collective repair – and received a special mention in the biennale awards. The restoration works, open from Tuesday to Friday, are being carried out by local artisans and specialised restorers with expertise in recovering stone, marble, terracotta, mural and canvas painting, stucco, wood and metal artworks. The beauty, however, lies in the photogenic fabrics, lit by a warm yellow glow, hanging from the walls within, gently wrapping the building’s surfaces, leaving openings that allow movement and offer glimpses of the ongoing restoration. Mobile scaffolding, used to support the works, also doubles up as furniture, providing space for equipment and subdividing the interior. Togo Pavilion (photography: Andrea Avezzù) Togo Pavilion The Republic of Togo has presented its first pavilion ever at the biennale this year with the project Considering Togo’s Architectural Heritage, which sits intriguingly at the back of a second-hand furniture shop. The inaugural pavilion is curated by Lomé and Berlin-based Studio NEiDA and is in Venice’s Squero Castello. Exploring Togo’s architectural narratives from the early 20th century, and key ongoing restoration efforts, it documents key examples of the west African country’s heritage, highlighting both traditional and more modern building techniques – from Nôk cave dwellings to Afro-Brazilian architecture developed by freed slaves to post-independence Modernist buildings. Some buildings showcased are in disrepair, despite most of the modern structures remaining in use today, including Hotel de la Paix and the Bourse du Travail, suggestive of a future of repair and celebration. Estonian Pavilion (photography: Joosep Kivimäe) Estonia Pavilion Another firm favourite this year is the Estonian exhibition on Riva dei Sette Martiri on the waterfront between Corso Garibaldi and the Giardini.  The Guardian’s Olly Wainwright said that outside the Giardini, it packed ‘the most powerful punch of all.’ Simple and effective, Let Me Warm You, curated by trio of architects Keiti Lige, Elina Liiva and Helena Männa, asks whether current insulation-driven renovations are merely a ‘checkbox’ to meet European energy targets or ‘a real chance’ to enhance the spatial and social quality of mass housing. The façade of the historic Venetian palazzetto in which it is housed is clad with fibre-cement insulation panels in the same process used in Estonia itself for its mass housing – a powerful visual statement showcasing a problematic disregard for the character and potential of typical habitable spaces. Inside, the ground floor is wrapped in plastic and exhibits how the dynamics between different stakeholders influence spatial solutions, including named stickers to encourage discussion among your peers. Venice Procuratie (photography: Mike Merkenschlager) SMAC (San Marco Art Centre) Timed to open to the public at the same time as the biennale, SMAC is a new permanent arts institution in Piazza San Marco, on the second floor of the Procuratie, which is owned by Generali. The exhibition space, open to the public for the first time in 500 years, comprises 16 galleries arranged along a continuous corridor stretching over 80m, recently restored by David Chipperfield Architects. Visitors can expect access through a private courtyard leading on to a monumental staircase and experience a typically sensitive Chipperfield restoration, which has revived the building’s original details: walls covered in a light grey Venetian marmorino made from crushed marble and floors of white terrazzo. During the summer, its inaugural programme features two solo exhibitions dedicated to Australian modern architect Harry Seidler and Korean landscape designer Jung Youngsun. Holcim's installation (photography: Celestia Studio) Holcim x Elemental Concrete manufacturer Holcim makes an appearance for a third time at Venice, this time partnering with Chilean Pritzker Prize-winning Alejandro Aravena’s practice Elemental – curator of the 2016 biennale – to launch a resilient housing prototype that follows on from the Norman Foster-designed Essential Homes Project. The ‘carbon-neutral’ structure incorporates Holcim’s range of low-carbon concrete ECOPact and is on display as part of the Time Space Existence exhibition organised by the European Cultural Centre in their gardens. It also applies Holcim’s ‘biochar’ technology for the first time, a concrete mix with 100 per cent recycled aggregates, in a full-scale Basic Services Unit. This follows an incremental design approach, which could entail fast and efficient construction via the provision of only essential housing components, and via self-build. The Next Earth at Palazzo Diedo (photography: Joan Porcel) The Next Earth At Palazzo Diedo’s incredible dedicated Berggruen Arts and Culture space, MIT’s department of architecture and think tank Antikythera (apparently taking its name from the first-known computer) have come together to create the exhibition The Next Earth: Computation, Crisis, Cosmology, which questions how philosophy and architecture must and can respond to various planet-wide crises. Antikythera’s The Noocene: Computation and Cosmology from Antikythera to AI looks at the evolution of ‘planetary computation’ as an ‘accidental’ megastructure through which systems, from the molecular to atmospheric scales, become both comprehensible and composable. What is actually on display is an architectural scale video monolith and short films on AI, astronomy and artificial life, as well as selected artefacts. MIT’s Climate Work: Un/Worlding the Planet features 37 works-in-progress, each looking at material supply chains, energy expenditure, modes of practice and deep-time perspectives. Take from it what you will. The 19th International Venice Architecture Biennale remains open until Sunday, 23 November 2025.

Missing the big picture US science is being wrecked, and its leadership is fighting the last war Facing an extreme budget, the National Academies hosted an event that ignored it. John Timmer – Jun 4, 2025 6:00 pm | 16 Credit: JHVE Photo Credit: JHVE Photo Story text Size Small Standard Large Width * Standard Wide Links Standard Orange * Subscribers only   Learn more WASHINGTON, DC—The general outline of the Trump administration's proposed 2026 budget was released a few weeks back, and it included massive cuts for most agencies, including every one that funds scientific research. Late last week, those agencies began releasing details of what the cuts would mean for the actual projects and people they support. And the results are as bad as the initial budget had suggested: one-of-a-kind scientific experiment facilities and hardware retired, massive cuts in supported scientists, and entire areas of research halted. And this comes in an environment where previously funded grants are being terminated, funding is being held up for ideological screening, and universities have been subject to arbitrary funding freezes. Collectively, things are heading for damage to US science that will take decades to recover from. It's a radical break from the trajectory science had been on. That's the environment that the US's National Academies of Science found itself in yesterday while hosting the State of the Science event in Washington, DC. It was an obvious opportunity for the nation's leading scientific organization to warn the nation of the consequences of the path that the current administration has been traveling. Instead, the event largely ignored the present to worry about a future that may never exist. The proposed cuts The top-line budget numbers proposed earlier indicated things would be bad: nearly 40 percent taken off the National Institutes of Health's budget, the National Science Foundation down by over half. But now, many of the details of what those cuts mean are becoming apparent. NASA's budget includes sharp cuts for planetary science, which would be cut in half and then stay flat for the rest of the decade, with the Mars Sample Return mission canceled. All other science budgets, including Earth Science and Astrophysics, take similar hits; one astronomer posted a graphic showing how many present and future missions that would mean. Active missions that have returned unprecedented data, like Juno and New Horizons, would go, as would two Mars orbiters. As described by Science magazine's news team, "The plans would also kill off nearly every major science mission the agency has not yet begun to build." A chart prepared by astronomer Laura Lopez showing just how many astrophysics missions will be cancelled. Credit: Laura Lopez The National Science Foundation, which funds much of the US's fundamental research, is also set for brutal cuts. Biology, engineering, and education will all be slashed by over 70 percent; computer science, math and physical science, and social and behavioral science will all see cuts of over 60 percent. International programs will take an 80 percent cut. The funding rate of grant proposals is expected to drop from 26 percent to just 7 percent, meaning the vast majority of grants submitted to the NSF will be a waste of time. The number of people involved in NSF-funded activities will drop from over 300,000 to just 90,000. Almost every program to broaden participation in science will be eliminated. As for specifics, they're equally grim. The fleet of research ships will essentially become someone else's problem: "The FY 2026 Budget Request will enable partial support of some ships." We've been able to better pin down the nature and location of gravitational wave events as detectors in Japan and Italy joined the original two LIGO detectors; the NSF will reverse that progress by shutting one of the LIGOs. The NSF's contributions to detectors at the Large Hadron Collider will be cut by over half, and one of the two very large telescopes it was helping fund will be cancelled (say goodbye to the Thirty Meter Telescope). "Access to the telescopes at Kitt Peak and Cerro Tololo will be phased out," and the NSF will transfer the facilities to other organizations. The Department of Health and Human Services has been less detailed about the specific cuts its divisions will see, largely focusing on the overall numbers, which are down considerably. The NIH, which is facing a cut of over 40 percent, will be reorganized, with its 19 institutes pared down to just eight. This will result in some odd pairings, such as the dental and eye institutes ending up in the same place; genomics and biomedical imaging will likewise end up under the same roof. Other groups like the Centers for Disease Control and Prevention and the Food and Drug Administration will also face major cuts. Issues go well beyond the core science agencies, as well. In the Department of Energy, funding for wind, solar, and renewable grid integration has been zeroed out, essentially ending all programs in this area. Hydrogen and fuel cells face a similar fate. Collectively, these had gotten over $600 billion dollars in 2024's budget. Other areas of science at the DOE, such as high-energy physics, fusion, and biology, receive relatively minor cuts that are largely in line with the ones faced by administration priorities like fossil and nuclear energy. Will this happen? It goes without saying that this would amount to an abandonment of US scientific leadership at a time when most estimates of China's research spending show it approaching US-like levels of support. Not only would it eliminate many key facilities, instruments, and institutions that have helped make the US a scientific powerhouse, but it would also block the development of newer and additional ones. The harms are so widespread that even topics that the administration claims are priorities would see severe cuts. And the damage is likely to last for generations, as support is cut at every stage of the educational pipeline that prepares people for STEM careers. This includes careers in high-tech industries, which may require relocation overseas due to a combination of staffing concerns and heightened immigration controls. That said, we've been here before in the first Trump administration, when budgets were proposed with potentially catastrophic implications for US science. But Congress limited the damage and maintained reasonably consistent budgets for most agencies. Can we expect that to happen again? So far, the signs are not especially promising. The House has largely adopted the Trump administration's budget priorities, despite the fact that the budget they pass turns its back on decades of supposed concerns about deficit spending. While the Senate has yet to take up the budget, it has also been very pliant during the second Trump administration, approving grossly unqualified cabinet picks such as Robert F. Kennedy Jr. All of which would seem to call for the leadership of US science organizations to press the case for the importance of science funding to the US, and highlight the damage that these cuts would cause. But, if yesterday's National Academies event is anything to judge by, the leadership is not especially interested. Altered states As the nation's premier science organization, and one that performs lots of analyses for the government, the National Academies would seem to be in a position to have its concerns taken seriously by members of Congress. And, given that the present and future of science in the US is being set by policy choices, a meeting entitled the State of the Science would seem like the obvious place to address those concerns. If so, it was not obvious to Marcia McNutt, the president of the NAS, who gave the presentation. She made some oblique references to current problems, saying, that “We are embarking on a radical new experiment in what conditions promote science leadership, with the US being the treatment group, and China as the control," and acknowledged that "uncertainties over the science budgets for next year, coupled with cancellations of billions of dollars of already hard-won research grants, is causing an exodus of researchers." But her primary focus was on the trends that have been operative in science funding and policy leading up to but excluding the second Trump administration. McNutt suggested this was needed to look beyond the next four years. However, that ignores the obvious fact that US science will be fundamentally different if the Trump administration can follow through on its plans and policies; the trends that have been present for the last two decades will be irrelevant. She was also remarkably selective about her avoidance of discussing Trump administration priorities. After noting that faculty surveys have suggested they spend roughly 40 percent of their time handling regulatory requirements, she twice mentioned that the administration's anti-regulatory stance could be a net positive here (once calling it "an opportunity to help"). Yet she neglected to note that many of the abandoned regulations represent a retreat from science-driven policy. McNutt also acknowledged the problem of science losing the bipartisan support it has enjoyed, as trust in scientists among US conservatives has been on a downward trend. But she suggested it was scientists' responsibility to fix the problem, even though it's largely the product of one party deciding it can gain partisan advantage by raising doubts about scientific findings in fields like climate change and vaccine safety. The panel discussion that came after largely followed McNutt's lead in avoiding any mention of the current threats to science. The lone exception was Heather Wilson, president of the University of Texas at El Paso and a former Republican member of the House of Representatives and Secretary of the Air Force during the first Trump administration. Wilson took direct aim at Trump's cuts to funding for underrepresented groups, arguing, "Talent is evenly distributed, but opportunity is not." After arguing that "the moral authority of science depends on the pursuit of truth," she highlighted the cancellation of grants that had been used to study diseases that are more prevalent in some ethnic groups, saying "that's not woke science—that's genetics." Wilson was clearly the exception, however, as the rest of the panel largely avoided direct mention of either the damage already done to US science funding or the impending catastrophe on the horizon. We've asked the National Academies' leadership a number of questions about how it perceives its role at a time when US science is clearly under threat. As of this article's publication, however, we have not received a response. At yesterday's event, however, only one person showed a clear sense of what they thought that role should be—Wilson again, whose strongest words were directed at the National Academies themselves, which she said should "do what you've done since Lincoln was president," and stand up for the truth. John Timmer Senior Science Editor John Timmer Senior Science Editor John is Ars Technica's science editor. He has a Bachelor of Arts in Biochemistry from Columbia University, and a Ph.D. in Molecular and Cell Biology from the University of California, Berkeley. When physically separated from his keyboard, he tends to seek out a bicycle, or a scenic location for communing with his hiking boots. 16 Comments

A Deadly Disease Is Eating Away at Caribbean Corals and Wreaking Havoc on Reefs. Could Probiotics Be the Solution? New research suggests the probiotic McH1-7 could help stop the spread of stony coral tissue loss disease among wild corals near Fort Lauderdale, Florida Scientists determined the most effective method of halting the disease was covering a coral colony with a weighted plastic bag, then injecting a seawater solution that contains the probiotic. They left the colony covered for two hours to allow the probiotic bacteria to colonize the coral. Hunter Noren Probiotics can be good for human health. Now, new research suggests they might also help protect coral reefs. A bacterial probiotic helped slow the advance of stony coral tissue loss disease—a fast-spreading and deadly condition—among wild corals in Florida, researchers report today in a new study published in the journal Frontiers in Marine Science. The probiotic may be a good alternative to antibiotics like amoxicillin, which temporarily curb the spread of the disease but must be reapplied frequently. In addition, scientists fear stony coral tissue loss disease may one day become resistant to these antibiotic treatments—just as “superbugs” that infect humans are building resistance to our own drugs. Antibiotics are meant to kill microorganisms, but probiotics are beneficial living microbes. The idea is that a probiotic can be incorporated into corals’ natural microbiomes, ideally offering them longer-lasting protection. First discovered in Florida in 2014, stony coral tissue loss disease attacks the soft tissue of more than 30 different species of coral. Without treatment, the disease eventually kills the corals, and their soft tissue falls off, revealing the white calcium carbonate skeleton below. In just weeks or months, it can devastate a whole colony. Stony coral tissue loss disease can be spread by fish that eat coral, as well as by boaters and divers who do not disinfect their gear. The condition has since expanded its range beyond Florida to reefs throughout the Caribbean. Several years ago, researchers looking at the great star coral (Montastraea cavernosa) discovered a probiotic called Pseudoalteromonas sp. strain McH1-7. Laboratory tests showed McH1-7 stopped or slowed the progression of stony coral tissue loss disease in infected corals. It also helped prevent the disease from spreading to healthy corals. But that was in the lab. Would McH1-7 be similarly effective in the ocean? Researchers were eager to find out, so they set up an experiment on a shallow reef off the coast of Fort Lauderdale. Study co-author Kelly Pitts, a research technician with the Smithsonian Marine Station, applies a paste containing the probiotic directly onto the disease lesion of an infected coral. Hunter Noren Experimenting with wild corals For the study, the scientists focused on 40 great star coral colonies that were showing symptoms of stony coral tissue loss disease. In one experimental condition, the researchers made a paste that contained McH1-7 and applied it directly onto the disease lesions. For comparison, they also applied the same paste, minus the probiotic, to some corals. In another condition, they covered infected coral colonies with weighted plastic bags, then filled the bags with seawater solutions made with and without McH1-7. They left the corals covered for two hours. “This created a little mini-aquarium that kept the probiotics around each coral colony,” says study co-author Valerie Paul, head scientist at the Smithsonian Marine Station at Fort Pierce, Florida, in a statement. The scientists completed all the treatments within the first 4.5 months of the project. Then, they returned periodically to gather tissue and mucus samples from the corals to measure changes to their microbiomes. Over the next 2.5 years, they took photos from a variety of different angles, which they then used to create 3D models that could track the disease’s progression. In the end, the results suggest covering the corals with plastic bags filled with the probiotic seawater solution was the most effective method. More than two years post-treatment, the colonies that received the probiotic bag had lost just 7 percent of their tissue, while colonies in the control bag condition faced 35 percent tissue loss. Scientists applied a probiotic paste directly to disease lesions on some corals. Kelly Pitts The probiotic paste, by contrast, appears to have made the situation worse: The corals that had the probiotic paste applied directly to their lesions lost more tissue than those treated with the control paste, which did not contain McH1-7. “We do not really know what is going on with the probiotic paste treatment,” Paul tells Smithsonian magazine in an email. But she has a few theories. It’s possible the high concentrations of McH1-7 contributed to localized hypoxia, or low-oxygen conditions that further harmed the already stressed corals, she says. Or, the probiotic could have changed the microbiome at the lesion site in some negative way. Another possibility is that McH1-7 produces antibiotics or other substances that were harmful at high concentrations. Amanda Alker, a marine microbiologist at the University of Rhode Island who was not involved with the study, wonders if this finding suggests McH1-7 is beneficial at specific dosages—a question future laboratory research might be able to answer, she tells Smithsonian magazine in an email. She’s also curious to know which specific molecular components of the probiotic are responsible for the increased tissue loss when applied as a paste. More broadly, Alker would like to see additional experiments validating the bag treatment method, but she says this “inventive” technique seems promising. “Their approach is a safer solution than antibiotic treatment methods that have been deployed to combat [stony coral tissue loss disease] in the field so far,” she says. “Further, this is a practical solution that could be implemented widely because it doesn’t require highly specialized equipment and has the ability to be used with any type of microbial solution.” Looking ahead to save reefs Probiotics are likely not a silver bullet for protecting corals. For one, researchers still don’t know exactly what causes stony coral tissue loss disease, which makes it difficult to determine how or why the probiotic works, Paul says. In addition, since the disease has spread to many different parts of the Caribbean, it might be challenging to use the bag treatment technique on all affected colonies. “We would need to develop better methods of deploying the probiotic through time release formulations or other ways to scale up treatments,” Paul says. “Right now, having divers swim around underwater with weighted bags is not a very scalable method.” The researchers have also conducted similar experiments on infected corals located farther south, in the Florida Keys. However, these tests have produced mixed results, probably because of regional differences in stony coral tissue loss disease. This is another hurdle scientists will likely need to overcome if they hope to expand the use of probiotics. “We probably need to develop different probiotics for different coral species and different regions of the Caribbean,” Paul says. Researchers returned to gather samples of tissues and mucus to see how the corals' microbiomes had changed. Hunter Noren Even so, scientists are heartened by the results of the experiments conducted near Fort Lauderdale. With more research, the findings suggest probiotics could be a promising tool for combatting the disease elsewhere. “Coral probiotics is a challenging field, because there are hundreds of different types of bacteria that associate with corals, and there are limitless experiments that need to be performed,” Amy Apprill, a marine chemist at Woods Hole Oceanographic Institution who was not involved with the research, tells Smithsonian magazine in an email. “These researchers made a major advance with their study by demonstrating the utility of whole colony treatment as well as the specific probiotic tested.” Apprill adds that, while antibiotics have been widely used to control stony coral tissue loss disease, scientists haven’t conducted much research to see how these treatments are affecting the plants and creatures that live nearby. “Using a naturally occurring bacterium for disease treatment may result in lessened impacts to other members of the coral reef ecosystem,” she says. Amid rising ocean temperatures, scientists expect to find even more diseased coral colonies in the future. Warmer waters may also allow other pathogens to thrive and proliferate. Against that backdrop, Apprill adds, probiotics and the different methods of applying them will be “major allies” in the fight to save coral reefs. Paul is also optimistic. Through research and field studies, she’s confident researchers will be able to develop interventions that can “help corals better survive changing environments and respond better to diseases and bleaching,” she says. Get the latest stories in your inbox every weekday.