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]

Macworld If you’re an Apple user we already know you like the finer things in life, and we’d put money on the fact your iPhone is controlling half the gadgets in your home. Why? Well, why not. It just works, and this seamless integration with smart home kit frees up your time and removes many of the frustrations of everyday life. But there is one smart gadget you might not have considered yet, and this robot can take care of the most gruelling chore of all: cleaning out the pool. Investing in a robot pool cleaner is a smart choice. Imagine being able to enjoy the luxury of a swimming pool, but with none of the hard work pool maintenance typically requires. The Dreame Z1 Pro is a great pick, unlocking plenty of extra free time to actually enjoy your pool. Imagine this: you’ve spent a very loooong week at the office, but now it’s Saturday, the weather has warmed up, and you can’t think of anything better than spending the weekend lounging in the pool you worked so hard to pay for. You grab a towel, get into your swimsuit, oil up… and then start picking out leaves that have fallen into the water. Oh, and here are some bugs, let’s get rid of those. And we’ll tackle that nice slimey film on top of the water. Did you scrub the pool walls and floors lately? Suddenly this luxurious dip you had been longing for feels like hassle. Let Dreame take the hard work out of pool maintenance It doesn’t have to be this way. Dreame can help, and the Z1 Pro ensures you enjoy hands-off pool maintenance anytime you need it. It deploys all of its 8,000GPH suction power to clean your pool top to bottom, picking up anything that’s fallen in the water and scrubbing the waterline. Much like a robot vacuum will tackle cleaning your floors, this one will stick to the bottom or sides of the pool and meticulously clean the surface, too. The robot’s sensors will quickly map out the pool while AI smarts divide it into areas so it can better tackle cleaning. The Z1 Pro even knows how to avoid obstacles such as drain covers, lights, and so on. This wireless robot pool cleaner can run for about 180 minutes on a single charge, cleaning up to 2,160 square feet in the process. That’s quite a lot of pool to cover! While it can normally do its own thing and clean wherever it deems necessary, you can also assume full control right from your iPhone or the bundled remote. Hands-free pool cleaning If you’re running an emergency clean just before guests arrive and you don’t want them to see the robot, use the remote to call it back to you. This LiFi-connected remote is suitable for fresh- and saltwater pools and can communicate with the robot even while it’s underwater, allowing you to assign tasks or steer the cleaner. Once the cleaning job is done, the robot will park itself at the edge of the pool so you can pick it up and set it up to recharge. Easy! To set up the robot and get cleaning reports, also install the Dreamehome app on your iPhone. You’ll see everything from the map to the multiple cleaning modes available, as well as the cleaning logs. What are you waiting for? Dreame’s Z1 Pro costs $1,499 at Dreame’s online shop, but right now it’s discounted to $1,099–and you can save a further 15% with the code PROMO15. We think you might just find the time saved on pool maintenance is worth more to you than the discount, however. Take a further 15% off Dreame Z1 Pro Use Promo Code PROMO15

The US is less than 250 years old, but some of its most important archaeological sites are older than the Viking seafarers, the Roman Empire, and the pyramids.Many help tell the story of how the first humans came to North America. It's still a mystery exactly how and when people arrived, though it's widely believed they crossed the Bering Strait at least 15,000 years ago."As we get further back in time, as we get populations that are smaller and smaller, finding these places and interpreting them becomes increasingly difficult," archaeologist Kenneth Feder told Business Insider. He's the author of "Ancient America: Fifty Archaeological Sites to See for Yourself."Some sites, like White Sands and Cooper's Ferry, have skeptics about the accuracy of their age. Still, they contribute to our understanding of some of the earliest Americans.Others are more recent and highlight the different cultures that were spreading around the country, with complex buildings and illuminating pictographs.Many of these places are open to the public, so you can see the US' ancient history for yourself. White Sands National Park, New Mexico Footprints at White Sands. National Park Service Prehistoric camels, mammoths, and giant sloths once roamed what's now New Mexico, when it was greener and damper.As the climate warmed around 11,000 years ago, the water of Lake Otero receded, revealing footprints of humans who lived among these extinct animals. Some even seemed to be following a sloth, offering a rare glimpse into ancient hunters' behavior.Recent research puts some of these fossilized footprints at between 21,000 and 23,000 years old. If the dates are accurate, the prints would predate other archaeological sites in the US, raising intriguing questions about who these people were and how they arrived in the Southwestern state."Where are they coming from?" Feder said. "They're not parachute dropping in New Mexico. They must have come from somewhere else, which means there are even older sites." Archaeologists simply haven't found them yet.While visitors can soak in the sight of the eponymous white sands, the footprints are currently off-limits. Meadowcroft Rockshelter, Pennsylvania The archeological dig at the Meadowcroft National Historic Site in 2013. AP Photo/Keith Srakocic In the 1970s, archaeologist James M. Adovasio sparked a controversy when he and his colleagues suggested stone tools and other artifacts found in southwestern Pennsylvania belonged to humans who had lived in the area 16,000 years ago.For decades, scientists had been finding evidence of human habitation that all seemed to be around 12,000 to 13,000 years old, belonging to the Clovis culture. They were long believed to have been the first to cross the Bering land bridge. Humans who arrived in North America before this group are often referred to as pre-Clovis.At the time, skeptics said that the radiocarbon dating evidence was flawed, AP News reported in 2016. In the years since, more sites that appear older than 13,000 years have been found across the US.Feder said Adovasio meticulously excavated the site, but there's still no clear consensus about the age of the oldest artifacts. Still, he said, "that site is absolutely a major, important, significant site." It helped archaeologists realize humans started arriving on the continent before the Clovis people.The dig itself is on display at the Heinz History Center, allowing visitors to see an excavation in person. Cooper's Ferry, Idaho Excavators at Cooper's Ferry in 2013. Loren Davis/Oregon State University One site that's added intriguing evidence to the pre-Clovis theory is located in western Idaho. Humans living there left stone tools and charred bones in a hearth between 14,000 and 16,000 years ago, according to radiocarbon dating. Other researchers put the dates closer to 11,500 years ago.These stemmed tools are different from the Clovis fluted projectiles, researchers wrote in a 2019 Science Advances paper.Some scientists think humans may have been traveling along the West Coast at this time, when huge ice sheets covered Alaska and Canada. "People using boats, using canoes could hop along that coast and end up in North America long before those glacial ice bodies decoupled," Feder said.Cooper's Ferry is located on traditional Nez Perce land, which the Bureau of Land Management holds in public ownership. Page-Ladson, Florida Divers search in the sediment at the Page-Ladson site. Texas A&M University via Getty Images In the early 1980s, former Navy SEAL Buddy Page alerted paleontologists and archaeologists to a sinkhole nicknamed "Booger Hole" in the Aucilla River. There, the researchers found mammoth and mastodon bones and stone tools.They also discovered a mastodon tusk with what appeared to be cut marks believed to be made by a tool. Other scientists have returned to the site more recently, bringing up more bones and tools. They used radiocarbon dating, which established the site as pre-Clovis."The stone tools and faunal remains at the site show that at 14,550 years ago, people knew how to find game, fresh water and material for making tools," Michael Waters, one of the researchers, said in a statement in 2016. "These people were well-adapted to this environment."Since the site is both underwater and on private property, it's not open to visitors. Paisley Caves, Oregon One of the Paisley Caves near Paisley, Oregon. AP Photo/Jeff Barnard Scientists study coprolites, or fossilized poop, to learn about the diets of long-dead animals. Mineralized waste can also reveal much more. In 2020, archaeologist Dennis Jenkins published a paper on coprolites from an Oregon cave that were over 14,000 years old.Radiocarbon dating gave the trace fossils' age, and genetic tests suggested they belonged to humans. Further analysis of coprolites added additional evidence that a group had been on the West Coast 1,000 years before the Clovis people arrived.Located in southcentral Oregon, the caves appear to be a piece of the puzzle indicating how humans spread throughout the continent thousands of years ago.The federal Bureau of Land Management owns the land where the caves are found, and they are listed on the National Register of Historic Places. Swan Point, Alaska Excavators working at the Swan Point site in June 2016. Charles Holmes/University of Alaska, Fairbanks Whenever people arrived in the Americas, they crossed from Siberia into Beringia, an area of land and sea between Russia and Canada and Alaska. Now it's covered in water, but there was once a land bridge connecting them.The site in Alaska with the oldest evidence of human habitation is Swan Point, in the state's eastern-central region. In addition to tools and hearths dating back 14,000 years, mammoth bones have been found there.Researchers think this area was a kind of seasonal hunting camp. As mammoths returned during certain times of the years, humans would track them and kill them, providing plentiful food for the hunter-gatherers.While Alaska may have a wealth of archaeological evidence of early Americans, it's also a difficult place to excavate. "Your digging season is very narrow, and it's expensive," Feder said. Some require a helicopter to reach, for example. Blackwater Draw, New Mexico A palaeontologist excavating a mammoth in Portales, New Mexico, circa 1960. Dick Kent/FPG/Archive Photos/Getty Images In 1929, 19-year-old James Ridgley Whiteman found mammoth bones along with fluted projectile points near Clovis, New Mexico. The Clovis people who made these tools were named for this site.Researchers studying the site began to realize the artifacts found at the site belonged to different cultures. Clovis points are typically larger than Folsom flutes, which were first found at another archaeological site in New Mexico.For decades after Whiteman's discovery, experts thought the Clovis people were the first to cross the Bering land bridge from Asia around 13,000 years ago. Estimates for humans' arrival is now thought to be at least 15,000 years ago.Eastern New Mexico University's Blackwater Draw Museum grants access to the archaeological site between April and October. Upper Sun River, Alaska Excavations at the Upward Sun River, Alaska. Ben Potter/University of Alaska, Fairbanks One reason the dates of human occupation in North America is so contentious is that very few ancient remains have been found. Among the oldest is a child from Upward Sun River, or Xaasaa Na', in Central Alaska.Archaeologists found the bones of the child in 2013. Local indigenous groups refer to her as Xach'itee'aanenh t'eede gay, or Sunrise Girl-Child. Genetic testing revealed the 11,300-year-old infant belonged to a previously unknown Native American population, the Ancient Beringians.Based on the child's genetic information, researchers learned that she was related to modern Native Americans but not directly. Their common ancestors started becoming genetically isolated 25,000 years ago before dividing into two groups after a few thousand years: the Ancient Berignians and the ancestors of modern Native Americans.According to this research, it's possible humans reached Alaska roughly 20,000 years ago. Poverty Point National Monument, Louisiana Poverty Point in Louisiana. National Park Service Stretching over 80 feet long and 5 feet tall, the rows of curved mounds of Poverty Point are a marvel when viewed from above. Over 3,000 years ago, hunter-gatherers constructed them out of tons of soil. Scientists aren't sure exactly why people built them, whether they were ceremonial or a display of status.The artifacts various groups left behind indicate the site was used off and on for hundreds of years and was a meeting point for trading. People brought tools and rocks from as far as 800 miles away. Remains of deer, fish, frogs, alligators, nuts, grapes, and other food have given archaeologists insights into their diets and daily lives.You can see the World Heritage Site for yourself year-round. Horseshoe Canyon, Utah The Great Gallery in Horseshoe Canyon. Neal Herbert/National Park Service Though remote, the multicolored walls of Horseshoe Canyon have long attracted visitors. Some of its artifacts date back to between 9,000 and 7,000 BCE, but its pictographs are more recent. Some tests date certain sections to around 2,000 to 900 years ago.The four galleries contain life-sized images of anthropomorphic figures and animals in what's known as the Barrier Canyon style. Much of this art is found in Utah, produced by the Desert Archaic culture.The pictographs may have spiritual and practical significance but also help capture a time when groups were meeting and mixing, according to the Natural History Museum of Utah.It's a difficult trek to get to the pictographs (and the NPS warns it can be dangerously hot in summer) but are amazing to view in person, Feder said. "These are creative geniuses," he said of the artists. Canyon de Chelly, Arizona The Antelope House at Canyon de Chelly National Monument. Michael Denson/National Park Service Situated in the Navajo Nation, Canyon de Chelly has gorgeous desert views and thousands of years of human history. Centuries ago, Ancestral Pueblo and Hopi groups planted crops, created pictographs, and built cliff dwellings.Over 900 years ago, Puebloan people constructed the White House, named for the hue of its clay. Its upper floors sit on a sandstone cliff, with a sheer drop outside the windows.Navajo people, also known as Diné, still live in Canyon de Chelly. Diné journalist Alastair Lee Bitsóí recently wrote about visiting some of the sacred and taboo areas. They include Tsé Yaa Kin, where archaeologists found human remains.In the 1860s, the US government forced 8,000 Navajo to relocate to Fort Sumner in New Mexico. The deadly journey is known as the "Long Walk." Eventually, they were able to return, though their homes and crops were destroyed.A hike to the White House is the only one open to the public without a Navajo guide or NPS ranger. Mesa Verde National Park, Colorado Visitors line up at Mesa Verde National Park. Shutterstock/Don Mammoser In the early 1900s, two women formed the Colorado Cliff Dwelling Association, hoping to preserve the ruins in the state's southwestern region. A few years later, President Theodore Roosevelt signed a bill designating Mesa Verde as the first national park meant to "preserve the works of man."Mesa Verde National Park holds hundreds of dwellings, including the sprawling Cliff Palace. It has over 100 rooms and nearly two dozen kivas, or ceremonial spaces.Using dendrochronology, or tree-ring dating, archaeologists learned when Ancestral Pueblo people built some of these structures and that they migrated out of the area by the 1300s.Feder said it's his favorite archaeological site he's visited. "You don't want to leave because you can't believe it's real," he said.Tourists can view many of these dwellings from the road, but some are also accessible after a bit of a hike. Some require extra tickets and can get crowded, Feder said. Cahokia, Illinois A mound at Cahokia in Illinois. Matt Gush/Shutterstock Cahokia has been called one of North America's first cities. Not far from present-day St. Louis, an estimated 10,000 to 20,000 people lived in dense settlements roughly 1,000 years ago. Important buildings sat atop large mounds, which the Mississippians built by hand, The Guardian reported.At the time, it was thriving with hunters, farmers, and artisans. "It's an agricultural civilization," Feder said. "It's a place where raw materials from a thousand miles away are coming in." Researchers have also found mass graves, potentially from human sacrifices.The inhabitants built circles of posts, which one archaeologist later referred to as "woodhenges," as a kind of calendar. At the solstices, the sun would rise or set aligned with different mounds.After a few hundred years, Cahokia's population declined and disappeared by 1350. Its largest mound remains, and some aspects have been reconstructed.While Cahokia is typically open to the public, parts are currently closed for renovations. Montezuma Castle, Arizona Montezuma Castle, a cliff dwelling, in Arizona. MyLoupe/Universal Images Group via Getty Images Perched on a limestone cliff in Camp Verde, Arizona, this site is an apartment, not a castle, and is unrelated to the Aztec ruler Montezuma.The Sinagua people engineered the five-story, 20-room building around 1100. It curves to follow the natural line of the cliff, which would have been more difficult than simply making a straight building, Feder said."These people were architects," he said. "They had a sense of beauty."The inhabitants were also practical, figuring out irrigation systems and construction techniques, like thick walls and shady spots, to help them survive the hot, dry climate.Feder said the dwelling is fairly accessible, with a short walk along a trail to view it, though visitors can't go inside the building itself.

The winners of the $TRUMP meme coin contest did get to see President Donald Trump speak at a private dinner closed to the press — but his speech was probably the least exciting part of their night. They did get a better, more valuable, and potentially more lucrative experience: the opportunity to network with the biggest crypto traders in the game, win watches worth hundreds of thousands of dollars, and attend a not-so-exclusive afterparty at the Capitol Hill Marriott afterward — all without having to complete particularly thorough background checks.The vetting process for entering the dinner was a ‘pretty light’ KYC checkAfter being whisked behind the gates of the Trump National Golf Club in Sterling, Virginia, past a throng of journalists snapping photos and protesters screaming at them for being corrupt, the 220 attendees went through security and had their IDs checked. According to one attendee, many were wealthy but some were living on normal-ish paychecks. The other guests, he said, were largely foreigners from overseas, all with an extremely high risk tolerance for gambling with crypto. The attendee said the vetting process for entering the dinner was a “pretty light” KYC check done by a third party, which he found odd considering he was about to have dinner in proximity to the president. (Know Your Customer is an anti-money-laundering compliance measure that banks, crypto exchanges, and other financial institutions are required to do.)“I talked to someone about getting into the White House, and it’s a lot more strict in terms of you have to show your passport and all that,” he told The Verge. “If this is true, it’s disappointing, but not surprising. When we sued during the first Trump administration to see who the Secret Service was running background checks on at Mar-a-Lago, we were told that the government wasn’t vetting the people meeting with Trump there, it was all done by his private business,” Jordan Libowitz, vice president of communications for Citizens for Responsibility and Ethics in Washington (CREW), wrote in email to The Verge. “This is a massive ethical issue that we reportedly have foreign nationals paying thousands if not millions of dollars to a sitting president to get access to him, and it’s all done in a way that the government does not have records of who they are. If you were drawing up a playbook for potential corruption, this is how you’d do it.” According to the attendee, Trump’s presence was limited to a speech from behind a podium bearing the presidential seal — despite the White House having previously called it a private event on the president’s “personal time” — then immediately flying back to Washington on Marine One. “For the most part, it was just him talking about his campaign, and about how he beat Biden, and blah, blah, blah, how we were in a terrible place with crypto before he got elected and now we’re in a great place,” the attendee said.Though Trump was greeted like a celebrity, with guests clamoring up close and hoping for signatures, the real draw of the event was Justin Sun, the crypto billionaire, who was swarmed with fans and selfie-seekers during the dinner. RelatedSun came in first place, having bought more than $20 million worth of $TRUMP during the contest. During a prize ceremony at the end of the night, he was presented with a $100,000 Trump-branded Tourbillon watch — the grand prize for the top four winners. (Unfortunately for third and fourth place, their Tourbillons were not ready by the time of the dinner and will be shipped to them later.) Later, there was a raffle for two other Trump-branded watches, each with an estimated retail value of $500. Other crypto stars were spotted at the event: Vincent Liu, chief investment officer at the Taiwan-based crypto trading firm Kronos Research; “Ice,” the founder of Memecore, a Singapore-based crypto organization that came in second place; and GAnt, a crypto influencer who came in fourth place and had been sharing his preparations for the dinner with his followers on Telegram. According to a report from the blockchain analysis company Nansen, the contestants collectively spent $394 million to participate in the dinner, with the winners spending anywhere between $55,000 and $37.7 million to participate.If people wanted to keep the festivities going after dinner, there were buses available to take them to the Capitol Hill Marriott back in DC, where Memecore was hosting a private afterparty at the rooftop bar. (The party had apparently been reserved for the top 25, but eventually it ballooned to include anyone else who wanted to attend.)The owners of $TRUMP and the White House have declined to publish a list of attendees, sparking outrage from lawmakers concerned about the potential for the token to be used for bribing the president. But several attendees were more than willing to make themselves known, giving on-the-record interviews with news outlets both before and after the dinner, posting photos and videos on their socials, and even openly discussing their experiences with their online Telegram followers. The dinner hosts themselves were just as eager to show off their success. A photographer was offering attendees the chance to find themselves in the event album via facial recognition. At the end of the night, after all the gifts had been handed out, Bill Zanker, the CEO of World Liberty Financial, asked everyone in the audience to put on their commemorative trucker hats, emblazoned with a slimmed-down Trump and the words “Fight! Fight! Fight!” for a celebratory photo. Everyone in the crowd flung theirs in the air, as if it were a college graduation. According to the attendee, Zanker then asked everyone to hashtag their photo with “Trump meme dinner or whatever” when they posted them. Although The Verge found photos of the hats on social media, we were not able to find any particular hashtag associated with them.The White House did not immediately respond to a request for comment.See More:

Sloths once came in a variety of sizes and lived in multiple settings in many parts of the world. A study in the journal Science examined sloth evolution over the past 35 million years, investigated multiple factors driving their growth and expansion throughout the world, and concluded that human hunting starting around 15,000 years ago drove their dramatic decline.Today, only six species within two genera remain. All are relatively small (especially compared to their largest ancestors) tree-dwellers that primarily live in the tropical rainforests of South and Central America.“These species are a tiny remnant of a once diverse American clade that was mostly made up of large-bodied species,” according to an editorial summary that accompanied the paper. Ancient Sloths Were Once WidespreadThat’s a huge contrast to sloth life during the late Cenozoic. During that period, more than 100 genera of sloths lived in a wide range of habitats and a variety of sizes, topping out at nearly 20 feet tall and weighing several tons.To investigate this diversity — and to track where, when, and why it collapsed — a team of scientists examined fossil measurements, DNA and protein sequences, and advanced evolutionary modeling. In doing so, they reconstructed sloth evolutionary history across 67 genera. They then investigated whether evolutionary changes in size were linked to habitat, diet, climate, predation, or other ecological pressures.Habitat Drove Sloth SizeThe findings show that habitat appeared to be a major driver in shaping their body size evolution. The earliest sloths were large and grazed on the ground. Some species adapted to tree dwelling and developed smaller body sizes. However shifts in both sloth size and dwelling didn’t happen in a straight line. The species size grew or shrunk as the climate warmed and cooled, and as ecosystems shifted from grasslands to woodlands.The species thrived for tens of millions of years, exhibiting the most variety in body sizes in the Pleistocene, which began about 2.6 million years ago.Ancient Humans Caused Dramatic DeclineThen, starting about 15,000 years ago, the creature experienced “a sudden and dramatic decline,” according to a press release.The researchers report that decline doesn’t mesh with any major known climate events. "Size disparity increased during the late Cenozoic climatic cooling, but paleoclimatic changes do not explain the rapid extinction of ground sloths that started approximately 15,000 years ago,” according to the paper. However, it does coincide with the expansion of humans into the Americas. The likely conclusion is that human hunting drove the extinction of the larger, ground-based sloths, while the smaller ones related to today’s creatures escaped by taking to the trees.Article SourcesOur writers at Discovermagazine.com use peer-reviewed studies and high-quality sources for our articles, and our editors review for scientific accuracy and editorial standards. Review the sources used below for this article:Before joining Discover Magazine, Paul Smaglik spent over 20 years as a science journalist, specializing in U.S. life science policy and global scientific career issues. He began his career in newspapers, but switched to scientific magazines. His work has appeared in publications including Science News, Science, Nature, and Scientific American.

Clownfish Shrink Down Their Bodies to Survive Ocean Heat Waves, New Study Suggests The adaptation appears to help the fish cope with high temperatures, since individuals and breeding pairs that shrank improved their survival odds Clownfish seem to become shorter during heat waves, according to the new study. Morgan Bennett-Smith A new study reveals that clownfish use a surprising strategy to adapt their bodies to ocean heat waves: They shrink. “[Clownfish] have these amazing abilities that we still don’t know all that much about,” says study co-author Theresa Rueger, a tropical marine ecologist at Newcastle University in England, to the Washington Post’s Dino Grandoni. The findings offer some hope for fish in the face of climate change, she adds. “There’s potential that maybe some other species will adapt in a way that will allow them to hang on longer than we think.” Rueger and her team didn’t initially plan to study a heat wave. They were monitoring how freshwater runoff might affect breeding clownfish in Papua New Guinea’s Kimbe Bay, when temperatures dramatically rose and warmed the water to 7.2 degrees Fahrenheit above average. But these conditions, they realized, offered a key opportunity for research. The scientists measured 134 clownfish in Kimbe Bay every month during the ocean heat wave, which spanned from February to August 2023. Astoundingly, 100 of those fish shrank. The researchers found that 71 percent of the dominant females and 79 percent of the breeding males reduced in size at least once over the study period. Their findings were published in the journal Science Advances on Wednesday. At first, lead author Melissa Versteeg, a PhD researcher at England’s Newcastle University, thought she was making a mistake in her measurements. She kept trying again. And again. “She had several people measuring them at the same time to really make sure that we’re confident with the numbers,” Rueger says to Melissa Hobson at National Geographic. But after these repeated attempts, she concluded the measurements were correct. The fish that shrank increased their chances of surviving the heat wave by 78 percent, according to the study. Some of the clownfish even shrank in pairs, reducing their size alongside their breeding partner—a move that also boosted their chance of survival. The study marks the first time a coral reef-dwelling fish has been documented to shrink in response to environmental and social cues, according to a statement from Newcastle University. A pair of clownfish swims near an anemone. When the studied fish became smaller, females maintained a larger size than males. Morgan Bennett-Smith Clownfish aren’t the only animals shifting their size because of heat. Fish around the world are adapting to warmer temperatures by downsizing their bodies. “This is another tool in the toolbox that fish are going to use to deal with a changing world,” says Simon Thorrold, an ocean ecologist at Woods Hole Oceanographic Institution who was not involved in the new work, to Adithi Ramakrishnan at the Associated Press. But these clownfish stand out from the rest. “Until now, when talking about shrinking fish, nearly all studies do not mean that fish literally shrink but that they grow to smaller sizes,” explains Asta Audzijonyte, a senior lecturer at the University of Tasmania in Australia who was not involved in the work, to the Washington Post. “This study, in contrast, reports observations of [clownfish] actually shrinking by a few percent of their total length over the course of a month.” Previous research has found that other animals, like birds and rodents, appear to have gotten smaller because of climate change. And marine iguanas will shrink in response to warmer water temperatures during El Niño years. The researchers don’t yet know how the clownfish are pulling off their shrinking act. One hypothesis is that the fish are reabsorbing their own bone matter, reports the Associated Press. They’re also not sure why, exactly, changing size is so advantageous to the clownfish. But it could be that a smaller size makes it easier to maintain oxygen levels or get by with less food available. “If you’re small, you obviously need less food, and you’re also more efficient in foraging a lot of the time,” explains Rueger to National Geographic. Still, this adaptation method can only go so far. The heat wave exacerbated coral bleaching, which decreases available reef habitat, and subsequent heat waves ultimately killed many of the fish the researchers studied. “We’ve lost many of those fish,” Rueger says to the Washington Post. Get the latest stories in your inbox every weekday.

It seems like a triumph for a cryptocurrency industry that has long sought mainstream acceptance: Top investors in one of President Donald Trump’s crypto projects invited to dine with him at his luxury golf club in Northern Virginia on the heels of the Senate advancing key pro-crypto legislation and while bitcoin prices soar.But Thursday night’s dinner for the 220 biggest investors in the $TRUMP meme coin has raised uncomfortable questions about potentially shadowy buyers using the anonymity of the internet to buy access to the president.While Democrats charge that Trump is using the power of the presidency to boost profits for his family business, even some pro-Trump crypto enthusiasts worry that the president’s push into meme coins isn’t helping their efforts to establish the credibility, stability and legitimacy they had thought his administration would bring to their businesses.After feeling unfairly targeted by the Biden administration, the industry has quickly become a dominant political force, donating huge sums to help Trump and crypto-friendly lawmakers. But that’s also served to tether the industry—sometimes uncomfortably—to a president who is using crypto as a platform to make money for his brand in unprecedented ways.“It’s distasteful and an unnecessary distraction,” said Nic Carter, a Trump supporter and partner at the crypto investment firm Castle Island Ventures, who said the president is “hugging us to death” with his private crypto businesses. “We would much rather that he passes common-sense legislation and leave it at that.” Concerns about Trump’s crypto ventures predate Inauguration Day At the swanky Crypto Ball held down the street from the White House three days before he took office on Jan. 20, Trump announced the creation of the meme coin $TRUMP as a way for his supporters to “have fun.”Meme coins are the crypto sector’s black sheep. They are often created as a joke, with no real utility and prone to extremely wild price swings that tend to enrich a small group of insiders at the expense of less sophisticated investors.The president’s meme coin is different, however, and has a clear utility: access to Trump. The top 25 investors of $TRUMP are set to attend a private reception with the president Thursday, with the top four getting $100,000 crypto-themed and Trump-branded watches.Trump’s meme coin saw an initial spike in value, followed by a steep drop. The price saw a significant increase after the dinner contest was announced. Its creators, which include an entity controlled by the Trump Organization, have made hundreds of millions of dollars by collecting fees on trades.First lady Melania Trump has her own meme coin, and Trump’s sons, Eric and Don Jr.—who are running the Trump Organization while their father is president—announced they are partnering with an existing firm to create a crypto mining company.The Trump family also holds about a 60% stake in World Liberty Financial, a crypto project that provides yet another avenue where investors are buying in and enriching the president’s relatives. World Liberty has launched its own stablecoin, USD1. The project got a boost recently when World Liberty announced an investment fund in the United Arab Emirates would be using $2 billion worth of USD1 to purchase a stake in Binance, the world’s largest cryptocurrency exchange.A rapidly growing form of crypto, stablecoins have values pegged to fixed assets like the U.S. dollar. Issuers profit by collecting the interest on the Treasury bonds and other assets used to back the stablecoins.Crypto is now one of the most significant sources of the Trump family’s wealth.“He’s becoming a salesman-in-chief,” said James Thurber, an American University professor emeritus who has long studied and taught about corruption around the world. “It allows for foreign influence easily. It allows for crypto lobbying going on at this dinner, and other ways. It allows for huge conflicts of interest.” How Trump changed his mind on crypto “I’m a big crypto fan,” Trump told reporters aboard Air Force One during last week’s trip to the Middle East. “I’ve been that from the beginning, right from the campaign.”That wasn’t always true. During his first term, Trump posted in July 2019 that cryptocurrencies were “not money” and had value that was “highly volatile and based on thin air.”“Unregulated Crypto Assets can facilitate unlawful behavior, including drug trade,” he added then. Even after leaving office in 2021, Trump told Fox Business Network that bitcoin, the world’s most popular cryptocurrency, “seems like a scam.”Trump began to shift during a crypto event at his Mar-a-Lago club in Florida in May 2024, receiving assurances that industry backers would spend lavishly to get him reelected. Another major milestone came last June, when Trump attended a high-dollar fundraiser at the San Francisco home of David Sacks.He further warmed to the industry weeks later, when Trump met at Mar-a-Lago with bitcoin miners. The following month, he addressed a major crypto conference in Nashville, promising to make the U.S. the “crypto capital of the planet.”Those close to Trump, including his sons and billionaire Elon Musk, helped further push his embrace of the industry. Sacks is now the Trump administration’s crypto czar, and many Cabinet members—including Commerce Secretary Howard Lutnick and Defense Secretary Pete Hegseth—have long been enthusiastic crypto boosters.“I don’t have faith in the dollar,” Transportation Secretary Sean Duffy said in a 2023 interview. “I’m bullish on bitcoin.” Trump + crypto: A political marriage of convenience Many top crypto backers were naturally wary of traditional politics, but gravitated toward Trump last year. They bristled at Democratic President Joe Biden ‘s Securities and Exchange Commission aggressively bringing civil suits against several major crypto companies.Since Trump took office, many such cases have been dropped or paused, including one alleging that Justin Sun, a China-born crypto entrepreneur, and his company engaged in market manipulation and paid celebrities for undisclosed promotions.Sun, who once paid $6.2 million for a piece of art involving a banana taped to a wall, and then ate the banana, helped the Trumps start World Liberty Financial with an early $75 million investment.Sun has disclosed on social media that he is the biggest holder of $TRUMP meme coins and is attending Thursday’s dinner.“I’m excited to connect with everyone, talk crypto, and discuss the future of our industry,” Sun said. Are Trump family profits hurting other crypto investors? Trump has signed executive orders promoting the industry, including calls to create a government bitcoin reserve. In March, Trump convened the first cryptocurrency summit at the White House.But some of the industry’s biggest names, often brash and outspoken, have kept mostly mum on Trump’s meme coins and other projects.“It’s not my place to really comment on President Trump’s activity,” Coinbase CEO Brian Armstrong said at a recent public event.Meanwhile, a top legislative priority for crypto-backers, a bill clarifying how digital assets are to be regulated, has advanced in the Senate. But some Democrats have tried to stall other pro-crypto legislation over the president’s personal dealings.“Never in American history has a sitting president so blatantly violated the ethics laws,” Democratic Rep. Stephen Lynch of Massachusetts said during a contentious House hearing earlier this month.The White House referred questions about dinner attendees to the Trump Organization, which didn’t provide a list of who is coming.“The President is working to secure GOOD deals for the American people, not for himself,” White House spokeswoman Anna Kelly said in a statement.In addition to Sun, however, some attendees have publicized qualifying for the dinner. Another will be Sheldon Xia, the founder of a cryptocurrency exchange called BitMart that’s registered in the Cayman Islands.“Proud to support President Trump’s pro-crypto vision.” Xia wrote in both English and Chinese on social media.Thurber, the expert on government and ethics, said Trump’s “personal attention to crypto at this dinner helps the crypto industry.”“But also it’s risky,” he said, “because they could all lose a lot of money.” —Will Weissert and Alan Suderman, Associated Press

Sick of warmed-over nostalgia? Here's how Disney could fix it.