TELL US ABOUT TOMORROW: UNDERSTANDING THE COMPLEX FORCES SHAPING THE FUTURE
AHMAD RASHAD: All right. Welcome back. Now remember when I said earlier that these Tech Summits are an early look at the future. This next panel embodies that spirit. It's going to be a wide-ranging discussion on the tech and business trends shaping tomorrow, and here to moderate is a journalist who many of you know very well, CNBC's senior media and technology reporter.
She covers all things content, distribution, and advertising. She also leads CNBC's Disruptor 50 coverage, showcasing the startups that are shaping the future. Please help me welcome Julia Boorstin.
(Applause.)
JULIA BOORSTIN: Thank you so much. I'm so thrilled to be here. This is really an honor, and I think we really have such an exciting panel.
And to start this conversation, I want to reference an old adage, which is that the best way to predict the future is to create it. It's been attributed to many people, including Abraham Lincoln.
But that idea that we are creating our future is really what this panel is about. And we're joined today by a group of people who are at the cutting edge of creating all of the amazing things that technology is driving for the near and far future.
So I'm going to introduce our panelists, starting with Connor Hayes, vice president and head of GenAI Products at Meta on the far end. Then we have Casey Wasserman, chairman and chief executive officer of Wasserman and chairperson and president of the LA28 Olympic and Paralympic Games, which we are very excited about.
And here to my immediate right is Scott Guthrie, executive vice president of Cloud and AI at Microsoft. Kevin Weil, chief product officer at OpenAI. Anne Wojcicki, co-founder and chief executive officer of 23andMe. And at the end, last but not least, Dr. Vasi Philomin, vice president of generative AI at Amazon Web Services.
So the goal of this panel is to get a vision of the future, and we're going to start really big picture. We'll eventually talk about the impact of AI in sports and all these things, but we have an opportunity now to really see what these amazing technologists are doing.
So, Scott, I'm going to kick things off with you because you lead a group of innovators at Microsoft. What are the emerging technologies that you're watching right now that could have a huge impact on our lives, on business and society?
SCOTT GUTHRIE: I think the thing that we're at Microsoft, but I think broadly in the industry, most excited about right now is obviously AI and generative AI.
And I think we're seeing several things, obviously. One is just the AI getting richer. And then I think the thing that gets us most excited is just you start to think about how is it going to be applied.
And I kind of use the analogy like when the iPhone first came out, it was a great device by itself, but it also, two or three years later, spawned whole new categories of experiences, whether it's an Uber or whether it's DoorDash or, you know, new types of scenarios that just weren't possible before.
And I think we're going to see the same thing with AI over the next two to three years, and it's just going to become something that's immersive in everything that we do, both in our lives but also across every business.
JULIA BOORSTIN: So not just integrated into existing businesses, but creating platforms for new types of businesses.
SCOTT GUTHRIE: New types of businesses that just weren't possible before.
JULIA BOORSTIN: So is there a particular industry or sector that you think will be most impacted by AI?
SCOTT GUTHRIE: I think you're going to see healthcare ultimately get transformed in a big way. I think that's an industry that has, I think, a big need to get more better solutions.
And so I was talking with Anne earlier about some of those. I think certainly anything that's kind of, I'll call it, a knowledge business, whether it's financial services, whether it's retail, I think that also is going to be completely transformed. And then ultimately I think some of the consumer experiences, what happens when you have a personal agent in your pocket that you can rely on for your day-to-day lives.
And those are the ones where I'm sure there's going to be things that will look obvious in three years that are not obvious to most of us in this room today.
JULIA BOORSTIN: Yeah. Well, we are all familiar with ChatGPT in terms of that personal experience. So it's a good segue over to Kevin, who I first interviewed maybe 15 years ago when you were at Twitter. So many jobs ago you were at Twitter, then you were running product at Instagram. You've been leading these various technological waves. And, of course, we all know ChatGPT.
But what are the perhaps lesser-known technologies that you're excited about right now that will be the use cases in the way that we're experiencing AI?
KEVIN WEIL: Yeah. Well, I'm curious: Who uses ChatGPT? Okay, pretty good. So Ruth was talking about Waymo, I think, just up onstage a little bit before. Have people ridden in a Waymo? You totally should if you haven't while you're here in San Francisco.
JULIA BOORSTIN: Actually a pretty low number of people have ridden in Waymo.
KEVIN WEIL: Yeah. So my experience in a Waymo, the first time, was the first, like, 15 seconds you're riding in this thing, you're like, oh, my god, watch out for that bike. And then five minutes later, you're like, oh, my god, this is amazing. This thing is just driving me around the city. This is the future. And then 10 minutes later, you're bored, looking at emails on your phone.
There's something -- I mean, ChatGPT is only two years old. And, you know, more or less the whole room raised their hand.
So on the one hand, it's incredible how quickly we as humans adapt to new technology and can make these things incredibly powerful tools that make us more efficient, you know, remove a bunch of busywork from our lives.
And then at the same time, they're getting better faster and faster and faster. So we went from GPT-3 to 3.5 to 4. And maybe, like, it took us six to nine months between those iterations. It's now taking us, like, two or three months between iterations in these reasoning models.
And so to answer your question, GPT-4 is a good engineer. It powers GitHub Copilot, powers a bunch of other things.
When we launched our first reasoning model, which is a model that can actually think, develop hypotheses, you know, refute its hypotheses, the same way that if I gave you a sudoku or a crossword puzzle or any kind of scientific research, you wouldn't just answer, you would think.
So the first version of these, kind of preview version, was like the millionth best engineer in the world based on competitive programming. And, you know, millionth engineer, okay, maybe. Fine, that's decent. It's like the top 2 or 3 percent of engineers in the world.
The next version, which was two months later, was like the thousandth best engineer in the world.
The model that we're going to launch soon is the 175th best engineer in the world.
And so we're going from, you know, 70 years ago, computers became better than humans at multiplying, at doing, like, basic computation. Fifteen years ago, computers were better, became better than humans at chess, and we've never looked back.
This is the year that computers are better than humans at writing code, writing software, and we will never look back.
And that's profound because software is at the heart of everything. It's not just engineering. It's if you can create software, if anybody in this room could, using English, create any kind of experience they wanted to using software, the world is a different place.
And that's one of the things I'm most excited about over the next, you know -- it's going to happen this year, and I think it's going to impact us all for the next three, five, ten years.
JULIA BOORSTIN: There are so many questions I could ask about implications for education, for upscaling, but that's a whole 'nother panel. So I'm going to hold on to those for later, but certainly raises some big, big, exciting questions and issues.
So, Anne, going to you next, you know, you've been working with AI for decades. Now you're using AI for drug discovery. How do you think AI is going to impact the way we approach our own personal genetic data and what it could be used for in the future?
ANNE WOJCICKI: Definitely. I think all of us -- I love, actually, explaining to my children in this way, where I have to go back in time and say, you know, back when I was little and I wanted to book a hotel in Paris, I had to wake up in the middle of the night. I had to call them, get on the phone, speak broken French, try to book it. I was like, You guys now just go online and you can get this, and it's all automated.
Health care is the only industry where it's really still dependent upon a one-to-one interaction with people. Like, if I want to get a blood test, I have to still call my doctor. If I want -- if I have a question, I have to go and call them to get the answers and follow up. Health care is very much of a protected industry where it's still based on the one-to-one.
And I think the real revolutionary opportunity is to be able to really have that co-pilot physician, wellness coach, that's going to be with you, that's going to really understand you, can pull in all of your wearable data, can pull in your genetics, pull in your medical record, and can actually start to coach.
And you see this now where ChatGPT has, congratulations, passed the MCAT, is actually starting to perform better. There was an interesting study about cardiology, about how training ChatGPT or any of the models with a primary care physician can make them on par with a specialist cardiologist.
So I think there's a lot of opportunities in the future for us to be able to have much better health care for ourselves and have this pilot and then go to the clinician or health care provider when you really need them, when you need them to touch you, you need something that's more invasive.
The second thing that is interesting, health care is a very fragmented industry with respect to the data. And I think that's what I personally am most excited about, is this opportunity for us to pull in -- through 23andMe, we have, you know, over 15 million customers. 85 percent have consented for research. There's companies out there like Datavant and others where I can now tokenize, I can pull in medical records, I can pull in all kinds of other data sets.
And what's going to be exciting is the ability for us to predict what is that next likely event. And the exciting thing is if you know what that next likely event is, then you could potentially prevent it.
And so it also has implications, as we talked about, for drug discovery, clinical trials. If you know who's likely to have an event, like a heart attack or develop fatty liver disease, something like that, then you can enroll those people in a study, and you're more likely to make that study successful.
So drug discovery is another area people talk a lot about. I think that the early part of drug discovery, meaning how you're developing a molecule, coming up with new ways to potentially drug what's known as an undruggable target, all of that is really exciting.
The challenge in healthcare is, I always like to remind people, you still have to put a drug through humans in clinical trials. So we can speed up a lot on the very early side, but it's not going to be an overnight revolution on the drug discovery side. But I think that will come eventually.
But I do think on care for all of us, that ability for us to have a co-pilot who really knows you, knows all your health information and can help guide you and give you suggestions and answer your questions that are going to be unique to you, I think is actually a real reality.
KEVIN WEIL: One thing. You both mentioned healthcare, just like a personal story. My son went and had a small surgery, you know, mostly not a big deal, but a small chance that there was something really serious. And they had to -- they went and, like, biopsied the thing, and I get a letter in the mail before I get a chance to talk to my doctor. And it's all these words that look really scary to me.
And before I had a chance to talk to my doctor because it's -- you know, even for those of us in this room who are incredibly lucky to be in the positions we're in, it was still hard to get ahold of my doctor. I just fed it to ChatGPT and said, well, what does this all mean. And fortunately it said, oh, you don't have anything to worry about, you know, this is what it means.
But that's like us in this room. Imagine 8 billion people around the world that probably have far less access to medical advice. And suddenly you're starting to have these free tools that, well, they're not perfect, but they're probably a lot -- I mean, they can be really meaningful when you need them.
ANNE WOJCICKI: There's a lot of cases now of people taking complicated medical situations and putting in ChatGPT and getting novel insights and actually that being the right path.
So I think there's a lot of opportunity. I would encourage everyone to start to experiment. Like, putting in your healthcare information is really interesting to see what then it spits out. And in a lot of cases, it's helping people resolve complicated situations.
JULIA BOORSTIN: And truly has potential to be a democratizing force in terms of giving access to healthcare.
So I want to shift gears a little bit away from this conversation that we could spend an hour just on healthcare to bring in Casey because you are not a technologist, you are in the entertainment and sports space. But you leverage technology and you're increasingly investing in tech tools for your athletes, you have this platform that you're building out.
How do you think about using technology as a game changer? No pun intended.
CASEY WASSERMAN: Well, thank you for making it clear to everyone in here I'm the least qualified person to be on this panel.
(Laughter.)
Which I agree with. And to Anne's point, I hope Dr. ChatGPT is much better for neurotic Jews like me than Dr. Google.
(Laughter.)
Because pretty much we're all dying of everything every time you go to Dr. Google.
Yeah, look, I mean, all of this technology, if it doesn't -- at least in our business, if it doesn't make the work we do for the clients we serve better, more valuable, or more relevant, more meaningful, then we're not really doing our job. And to have all these tools now are really extraordinary.
So, you know, one of the things we've been doing is how do you talk about value of talent. There used to be Q scores, or there are Q scores, which are kind of this old-and-cold data. It's like the talent version of the Nielsen ratings, which is 2,000 people have boxes in their homes, which somehow still exists.
And how do you take realtime data, performance data, earned media, paid media, and create what we're calling sort of a relevancy index, and in real time for all athletes, to make it easier for brands to connect and consumers to connect with talent and find value in a way that is more meaningful and more directly relevant?
Think about our music business. Our music business, we're a music agent, we booked 75,000 shows last year around the world. We have touring data, we have access to streaming data, we have things like weather that impact attendance, all these things.
How can we be better at pricing, touring, routing, sizing venues for our talent and add -- you add a couple thousand dollars a night of gross. You're adding hundreds of millions of dollars for the same amount of work in the same exact places without doing anything else. And that's driven by technology.
And so all of those things are really valuable. We represent Seth Gordon, who's a producer and a writer. He wrote "Back in Action" and also "The Recruit," both number one shows on Netflix.
But "Back in Action," the movie with Jamie Foxx, when they were filming it, he had to take a break during COVID. And one of the kids in the movie grew.
JULIA BOORSTIN: You can't stop that.
CASEY WASSERMAN: So they used AI to take the voice back so that the voice was a 9-year-old, not a 12-year-old, because the voice changed. And they used AI trained on the original filming so that when the 12-year-old spoke, it still sounded like the 9-year-old.
That made his process better. He can in real time experiment with what fans are reacting to, as opposed to a test screening, now in real time on these streaming platforms.
So to take all that data and information and bring value to it and actually create different outputs of product is a really powerful opportunity.
JULIA BOORSTIN: And is there a certain technology that -- I mean, you look at this GenAI, you know, content, we'll talk about Sora in a little bit -- that you think that this is going to be transformative for your business? Or is it more just the ability to leverage data in ways you couldn't before?
CASEY WASSERMAN: Well, look, my concern for all of this is I actually believe that copyrights and intellectual property matter. So let's not be confused. I mean, the T-shirts that I agree with what they were saying, that all the celebrities saying a certain thing to a certain former rapper about his antics, but those were all AI driven.
So you had this weird thing where they believed in what they were saying, but they used their people illegally. There are all these music platforms that are saying write a song like Kendrick Lamar, who we represent. Well, isn't that actually just stealing Kendrick Lamar's intellectual property? Where is that line?
So that's the stuff that really keeps me up at night, and how do you protect those things?
There could be a lot of upside to those things, too. I'm not minimizing that, and we always try and be thoughtful and entrepreneurial.
Look, the good thing for us is we don't -- that's why I'm on this panel, because I'm not the smart one -- we don't have to be smart enough to know where the world is going. We happen to be in a business that will almost disproportionately benefit from where the world is going.
JULIA BOORSTIN: Yeah.
And, Connor, you've been at Meta for -- previously Facebook, for 15 years, you just said. And your job is to make bets that are going to be where the world is going.
I was just backstage telling Connor how much I love trying the Orion glasses, these augmented reality glasses, which were truly an unparalleled experience.
And these are a long-term bet. It'll be a couple years before I can go buy a pair and then watch an NBA game through them. How do you think about making a bet for 10 years out?
CONNOR HAYES: I mean, the way that we think about this is just imagining the future in a way that helps people connect in a better way. I mean, Meta has always been a company that's about driving connection between people.
In the very beginning, it's like you and your friends. I think increasingly it's how do you experience the world around you and your interests.
That's why we've been investing in AR and VR. We've had a big year over the course of last year with the Meta Ray-Ban glasses, which have become pretty popular, and those are a vehicle for people to interact with AI as well.
But all the things that folks are talking about here, what we're most excited about is as model capabilities get better, as the application ecosystem built on top of it gets more robust, it will be increasingly possible for you to go about your day and experience the world with an assistive AI on your face at all times. And that can be a really powerful thing too.
In the modeling space, there's this sort of future domain that we, at least inside FAIR at Meta, call world models. This is like instead of taking data from textbooks and encyclopedias and processing that with a machine and trying to build intelligence, it's taking data from physical spaces, like how far away is the front row for me right now? How are my hands and fingers moving? What is this rug made out of? Is it soft or hard?
And as computers can start to understand those things, as you navigate the world, we can give you predictive insights about what's happening or about to happen.
I mean, Waymo, I think, is a good example of how this works really well today. I saw a video recently of a woman falling off of a bicycle in front of a Waymo, and it veers out of the way, like instantly, where a human driver probably wouldn't have done the same thing.
So imagine that same capability that the Waymo had on your face, instructing you about how to go about your day and live your life. Those things can get really exciting.
I also think there's an entertainment angle to all of this as well. I mean, Julia, we were -- with Orion, you know, I've been meeting more and more people who have demoed them. Casey, I think you tried them too. The pass-through of having glasses on your face and the ability with a sensor on your wrist to manage apps and visually control what's in front of you is pretty fascinating.
My dream for that is like you're sitting in an NBA arena and able to use the glasses to, you know, like tap in on a player and see what their stats are for the day.
You know, we're starting small. We actually have some stuff this weekend. We're going to be streaming the Dunk Contest on Quest 3 live so you can get into an immersive view and see that.
But over time you'll start to see the intersection of the world around you and this technology that we're talking about, I think, happening through wearable devices, which is where we're really invested.
JULIA BOORSTIN: Yeah, and I'm having to ask one newsy question. There was a headline this morning saying that Meta is investing in building robots. What does this mean? Why are you investing in building robots?
CONNOR HAYES: The headline was that we're starting a robotics division, which is, you know -- I think Julia asked me this morning, like why Meta? And, you know, my response was, If I had told you ten years ago that we're doing glasses, you would have said, Why Meta?
This is an R&D project at this point. But, you know, similar to what I was just saying, as models begin to understand the world, and we're trying to drive value for you so that you can, you know, save time, be more entertained, be more connected with the people and things around you, you know, we believe that there may be something in the robotics space there if a robot can, you know, make your coffee for you in the morning before you even get up, or something like that.
You know, there's value to be created there, but it's still really early, and it's an early R&D team.
JULIA BOORSTIN: Humanoid robotic coffee maker machine.
DR. VASI PHILOMIN: I want to get in on that a little bit. I think one of the trends to watch out for, I think, for digital sports, it's the fusion of physical and digital sports. You can already see events there, out there, hybrid competitions like Games of the Future, that sort of exemplify the trend.
And so you sprinkle in a little bit of generative AI, you sprinkle in a little bit of new sensor technology, and then you're going to have a big impact on sports in general going forward.
JULIA BOORSTIN: I want to take a step back and talk more about your big-picture philosophy here, but tell us a little bit more about that. Is the fusion of physical and digital, is that for the consumption or for the playing?
DR. VASI PHILOMIN: It's for the consumption, was what I was talking about.
JULIA BOORSTIN: Yeah. And then, so you're saying this is the next generation of how we all watch?
DR. VASI PHILOMIN: Yeah, yeah. I think there's a lot that could happen there, and it's something to watch out for over the next 10 years or so.
JULIA BOORSTIN: And so, Vasi, Amazon supports so many different businesses, so many diverse sectors. Which industries are you seeing -- I'm sure that these are your customers -- that are leveraging AI in ways that are surprising or perhaps more valuable than you would have anticipated?
DR. VASI PHILOMIN: Yeah. 95 percent of the Fortune 500 are on AWS, and I think about 80 percent of the PitchBook unicorns are on AWS. So we see a whole bunch of different use cases across many, many different industries. And it's at different scales as well, you know, startups, there's small, medium businesses, and then there's large enterprises.
If you had asked me the question 18 months ago, who would be the industries that would adopt generative AI the fastest and see value in it, I wouldn't have guessed, actually, but we do have the answer now. It happens to be the fintech businesses as well as the healthcare/life sciences businesses that are adopting it the fastest.
And the reason why I said I wouldn't have guessed it is because these industries are typically conservative. They're a little bit slower in adopting new things. But we know the answer as to why, you know, why are they the fastest now. It's because they have the data strategy in place.
Because of the regulations that they've had to go through, they have their data in one place, which is it's easily discoverable and it's shareable at the right moment with the right applications and the right people. And because of that reason, I think you can get a lot of value with generative AI.
JULIA BOORSTIN: Do you think there's going to be a digital divide between the big guys, the pitchfork startups, and the small businesses, or do you think that AI is going to be more universally accessible?
DR. VASI PHILOMIN: Yeah, I definitely don't think there's going to be a digital divide because I think, for the first time, AI is truly accessible to every person.
In the past, if I just go through how I used to build systems -- let's just take a simple example. Let's say you want to translate English to Spanish. In order to do that, I would have to collect a whole bunch of data that have the pairs, you know, English-Spanish pairs. And then I would have to hire scientists that are hard to find and give them six to eight months.
And at the end of the period, they're going to come up with a model that can do the translation, right? And if I now want to do something else, like summarization, I have to go through the same process all over again.
But now, with the models that have seen web-scale data, they are very generalizable. So you can take one off the shelf, and you can simply give it examples of inputs and outputs. And most of the times you don't even have to do that. They're going to be able to do anything you want them to do.
So I recommend to all of the customers, stop trying to hire scientists. Just go and educate your line of business developer on the GenAI tools, and they're going to be able to innovate much faster than anybody else.
KEVIN WEIL: It resonates. I was in Tokyo and Seoul in the past couple weeks, and in a bunch of meetings where we had -- we had live translation, but then afterwards you're going and talking individually with people. And in the past I just wouldn't have been able to talk individually.
Now I went and put my phone down on the table in front of us, and I said, Hey, ChatGPT, when I say something in English, I want you to repeat it in Korean; and when you hear something in Korean, I want you to repeat it in English.
And I had live, business-level conversations with people with whom I did not share a single word in common or language in common.
And you can do that for free now. I mean, it's amazing. So I really think this stuff changes the world.
JULIA BOORSTIN: Yeah, and it sounds like also you're optimistic about the democratizing force, not just when it comes to healthcare issues.
Scott, how do you see these GenAI tools having an impact across your ecosystem in terms of the way people are working, communicating?
SCOTT GUTHRIE: Yeah, I mean, I think it is really democratizing this kind of amazing capability. And so, you know, we see it in, like, so many use cases. You know, sometimes they're profound. Sometimes they are just 20 percent better than the typical workflow. But those 20 percent wins add up to 80, 90 percent wins in a lot of cases.
And, you know, I think one of the ones that we're really excited about that we've seen both, you know, good uptake but also real satisfaction from is in the sort of patient-doctor conversations.
And so, you know, similar to your example, you think about, like, when you go to a physician today, often, because of insurance needs, you know, the physician is either typing notes furiously, not paying attention, or you have to have a human scribe in the room, which is sort of awkward, and it's needed to create that clinical documentation.
And now you can just take your cell phone out as the doctor, get permission, hit record, have the conversation, look at the patient, you know, really get to the insights of the problem, and then it'll create the clinical documentation. Doctor can review it, hit submit. It goes into the medical record system.
And, you know, we've just -- we get these sort of love notes from doctors who are just saying: I was about to quit medicine. I spent three hours a night having to catch up on notes. I hated my job. Thank you.
And I think that's also kind of one of the exciting things about AI, is how can we actually transform jobs and make them more satisfying.
Now, on the converse, there's also going to be a lot of disruption. And I think that is one of the big topics that will be, you know -- we will see more and more in the coming years is because there's both great use cases, new use cases, and it'll be very disruptive to, you know, acknowledge to some existing use cases that will impact people in businesses.
JULIA BOORSTIN: And this question of job replacement, you know, how and at what sort of timeline can you up-skill workers or retrain workers so they're able to do that? Do you see an immediate, sort of near-term workforce disruption coming?
SCOTT GUTHRIE: I think the place -- I mean, we'll probably see it. I think yes. I mean, I think the big question is going to be how much can we all, as workers, absorb the technology and adapt? And I'm kind of long-term, long on our ability to adapt and that we will be able to find new jobs.
You know, it's fascinating. There was a good jobs -- or an interesting jobs report that came out last week. And, you know, in general, unemployment in the U.S. is sort of in the 4 percent range. You know, for the first time ever, developers are now up to over 5 percent unemployment, which is -- I can't remember the last time unemployment in the developer base was higher than the national average.
Is that a blip? Is that AI? Is that because there's just a lot of people entering the workforce in that particular discipline? We'll see. And I'm not trying to project that that's going to go skyrocket.
JULIA BOORSTIN: Is that because of what Vasi was saying, is that maybe you don't need as many developers, you need people who have the creative skills who are going to be asking the AI to be their engineer?
SCOTT GUTHRIE: Or I think what's also going to happen is, as people embrace the AI technology, the developers become more valuable, and you actually, hopefully, need more developers in the world because they can now do more, have bigger impact. And, you know, we've seen with technology in the past, that's typically the trend.
JULIA BOORSTIN: Yeah.
SCOTT GUTHRIE: And so I'm still very optimistic.
JULIA BOORSTIN: You're thinking it's transitional.
SCOTT GUTHRIE: But I think, you know, we need to acknowledge it's going to be sometimes bumpy if you feel like your company or your industry is suddenly adopting technology and your exact day-to-day job is changing.
And, you know, it's a journey that we'll go through together, and but it is, you know -- I think we ought to be a little eyes wide open on it.
JULIA BOORSTIN: Yeah.
SCOTT GUTHRIE: At the same time, I'm super, super bullish, both as a consumer of it and then also the long-term benefit that I think it will bring.
JULIA BOORSTIN: Yeah.
Connor, you're nodding. And I also know that at Meta you have this very unique approach of being open source. And you just mentioned the stat of how many times Llama has been downloaded.
CONNOR HAYES: Over 800 million.
JULIA BOORSTIN: Over 800 million times, which really changes the equation in terms of what the infrastructure is going to be of so many of these different small businesses or Fortune 500 companies that are integrating Llama.
What does this all mean for the business opportunities for Meta down the road?
CONNOR HAYES: Yeah, I mean, for the -- so we've trained a model called Llama that we've been releasing open source, which means that anybody can go take the weights of the model, download them, use it. We work with companies like Microsoft and Amazon and others as cloud partners, where they also host it and can give people access to it.
I think one of the goals that we had when we started to open source Llama was to accelerate this democratization that we're all talking about. It sort of pushes the whole industry forward. It adds more competition into the space. Like, you see the technology advance faster when it's in the hands of more people.
But you also can start to standardize sort of the blocks on which the entire industry is building. And then when people make improvements upon the model that you've developed, like merging them back into the things that you're doing becomes easier to do, right?
So that's part of the philosophy for us. We feel like it's a win-win for Meta and for the ecosystem.
The big thing that we're focused on now, though, is like the way that we've open sourced Llama in the past I think has really advantaged people who know what they're doing. You have to patch together a few different tools usually to complete, like, a complex task. But we do see a lot of examples of people who know what they're doing having phenomenal results.
The future is for the less advanced folks, you're not a research scientist, you're like a line developer at a company, how can you take something like a Llama open source model out of the box and get value from it quickly?
I actually think this is something that, you know, like OpenAI has sort of established itself as like the AI shop for a lot of people, right? You show up there and you get AI. I don't think people go to Hugging Face and download Llama and they're like, all right, let's get going with AI. And we need to start pushing the toolchain to make it easier for people to ramp on.
So that's kind of how we're thinking about it. And, again, like, if you do that over time and more developers are relying on your model and they're asking you for more services, like, that just can become a business opportunity as well.
JULIA BOORSTIN: Yeah. So this really seems like consensus about the democratizing power, the quick-moving pace of it, and also the fact that these are going to be tools that are integrated into every single type of business, which is a good segue to get back to the sports of it all.
So, Anne, tell us a little bit about how your tools are being deployed for athletes.
ANNE WOJCICKI: Yeah. So for those of you who are 23andMe customers, people love to talk about our fast-twitching muscle report, which is if you -- they found a very high propensity, almost all elite athletes have this genetic variant of a fast-twitching muscle.
So people love to learn about their genetics and what their likelihood might be. So do they have the potential to be an elite athlete -- I do, but I'm clearly not -- or do you not?
So where we're really focusing now is on this world of risk prediction. So if you look around the room, and everyone can talk about their parents had a different condition or they have siblings that had a condition -- my sister just passed away of lung cancer -- we all wonder, like, what are your risks and what can you do?
So we're really focused on that area of risk prediction. And for athletes, you think about what makes an athlete so successful is in many ways also the ability to stay in the game and avoid injury.
So we think a lot about things like ACL tears, bone issues, like what are the ways -- right ways to train? What is sort of your peak way of training?
So we look at collecting that data from all of our customers. We run all kinds of surveys. And then we help people understand, like we generate reports, then, on things like ACL tears, osteoporosis, other areas that could then end up impacting athletes.
So we do look more and more the whole world of personalization in your health. So whether that's personalized nutrition, helping people understand how much they should sleep, helping people understand how much they should train, all of these data sets are now starting to be generated.
And this is where AI really comes in for helping us then predict really how this comes down specifically to you and how you should best train and how you can stay healthy.
JULIA BOORSTIN: Yeah, certainly something we're all thinking about.
We are almost out of time, but I want to get a couple quick final thoughts on sports.
Vasi, tell us a little bit about what Amazon is doing right now around sports entertainment consumption technology. You mentioned this combination of physical and digital, but what does this actually mean that Amazon is doing?
DR. VASI PHILOMIN: Yeah. We've got a lot of customers of AWS, large sports organizations, the NFL, Bundesliga, F1. They're all trying to use generative AI to completely reinvent their fan experience.
And the NFL was the group that started on this a while ago. And then working together with us, they came up with next-gen stats, which we all very well know now about. This allows broadcasters to kind of tell stories backed by data.
But I think the NFL is taking it to the next level now with their Digital Athlete program which focuses on player health and safety. They've even come up with new rules for the game. Like the dynamic kickoff that they've come up with is actually meant to reduce injuries during kickoff returns.
So I'm seeing that -- I'm definitely seeing a trend for a lot of these organizations in terms of reinventing that fan experience.
JULIA BOORSTIN: Well, I've asked you all to look into your crystal ball and tell us about the future. And the bottom line is that the data is what's going to lead us into the future.
I'm so sorry we're out of time. I could have kept going for another couple hours, but I know that you have a whole weekend. Thank you all so much, and thank you so much for being here.
(Applause.)
AHMAD RASHAD: Thank you very much, Julia.
We're going to take another break. Now, if you go down one level, there are boxed lunches available. Feel free to grab one and bring it back up here.
And the dog, you see the dog I was with early on, my dog, was in my dressing room? He's not in the dressing room now. So if you see him, run.
FastScripts Transcript by ASAP Sports