Seriously, What Is ‘Superintelligence'?
WIRED20-06-2025
In this episode of Uncanny Valley , we talk about Meta's recent investment in Scale AI and its move to build a superintelligence AI research lab. So we ask: What is superintelligence anyway? Meta AI at the Meta pavilion ahead of the World Economic Forum (WEF) in Davos, Switzerland, on Saturday, Jan. 19, 2025. Photo-Illustration: WIRED Staff; Photograph:Meta just announced a major move in its AI efforts—investing in Scale AI and building a superintelligence AI research lab. While Meta has been trying to keep up with big names in the AI race, such as OpenAI, Anthropic and Google, the company's new strategy includes dropping some serious cash to acquire talent and invest in Scale AI.
Today on the show, we dive into the deal between Meta and Scale AI, including what Meta aims to get out of investment, and we ask the question we are all wondering: What is superhuman intelligence, anyway?
You can follow Michael Calore on Bluesky at @snackfight, Lauren Goode on Bluesky at @laurengoode, and Katie Drummond on Bluesky at @katie-drummond. Write to us at uncannyvalley@wired.com. How to Listen
You can always listen to this week's podcast through the audio player on this page, but if you want to subscribe for free to get every episode, here's how:
If you're on an iPhone or iPad, open the app called Podcasts, or just tap this link. You can also download an app like Overcast or Pocket Casts and search for 'uncanny valley.' We're on Spotify too. Transcript
Note: This is an automated transcript, which may contain errors.
Michael Calore: How is everybody doing this week?
Katie Drummond: Well, I'm back and I'm so happy to be here. I ate so much butter in France last week, the first couple nights I had to myself. And so what would a normal person do in Paris? Maybe they'd go out and sit at the bar and have dinner alone, maybe they'd meet up with a friend. I ate butter and bread alone in my hotel room. And let me tell you, if you're listening out there and you are a mom and you have a young kid or young kids, if you're married, if you struggle with having a spouse and a child and maybe some pets and a busy job, there is no better experience than eating French butter and bread alone in a hotel room.
Michael Calore: Wow.
Lauren Goode: Uncanny Valley , brought to you by the dairy lobby.
Katie Drummond: By the French dairy lobby. I feel incredible. How are you guys?
Lauren Goode: I'm OK. I got bangs.
Katie Drummond: You did?
Lauren Goode: Yeah. So most of our listeners can't see it unless you watch our new video promos online, but I got bangs. There's often a correlation between things going on in the world and women cutting their bangs. That's all I'm going to say about that. But otherwise, I'm great. I'm great. Rate the bangs, go online. Thumbs up, thumbs down.
Katie Drummond: Five out of five.
Lauren Goode: Thank you. Katie, really, you were my inspiration here.
Katie Drummond: Oh, no, that's too kind. But I do love a bang and hate a forehead.
Michael Calore: Well, have I got a haircut for you.
Lauren Goode: Well, Mike, how are you doing? Sorry about that, Mike.
Katie Drummond: But there's just so much smarts in there, Mike. That's the thing.
Michael Calore: I'm just as God made me.
Katie Drummond: How are you doing?
Michael Calore: I'm doing great. I don't have any hair stories. I haven't eaten any butter recently, so I'm feeling really left out right now. This is WIRED's Uncanny Valley , a show about the people, power and influence of Silicon Valley. Today we're talking about Meta's recent investment in Scale AI, and its move to build a superintelligence AI research lab. It's the latest effort from Meta to compete with the big names in the AI race like OpenAI, Anthropic, and Google. But Meta is taking a different approach. Not only are its AI models open source, but in typical Meta fashion, it seems to be trying to outspend its competitors to acquire top talent, and its Scale AI investment, which is not an acquisition, is just part of that strategy. We'll dive into what Meta is hoping to get from this investment and what it's actually getting, and whether this move could give the company the competitive advantage it's seeking. Plus we will ask, what is superintelligence anyway? I'm Michael Calore, director of consumer tech and culture here at WIRED.
Lauren Goode: I'm Lauren Goode. I'm a senior correspondent at WIRED.
Katie Drummond: And I'm Katie Drummond, WIRED's global editorial director.
Michael Calore: So let's start off by diving into Scale AI. Unlike Meta, the company is not what you would call a household name anywhere outside of Silicon Valley, but it's certainly made a name for itself in the AI world. What's up with Scale?
Lauren Goode: Scale AI is a data labeling company. Sounds very sexy, doesn't it? They do the grunt work of analyzing and categorizing the data that is later distributed to larger AI models in order to train them. It's nothing fancy, but they kind of perform an essential function for machine learning programs to improve. And they have some pretty big customers. Companies like OpenAI and Google have been among their clients. Earlier this year, our colleague Will Knight reported on Scale AI's new platform that allows AI models to be automatically tested against key benchmarks and pinpoint any weaknesses in the models. So basically Scale AI has been making this concentrated effort to be a key partner as it's working with these larger AI companies.
Katie Drummond: I think it's also worth pointing out that really at the heart of the company's success is its founder, Alexandr Wang, without an E, right?
Michael Calore: Yeah.
Lauren Goode: Yeah.
Michael Calore: He's the Flickr of CEOs.
Katie Drummond: There you go. He's 28 years old, and at one point in time he was actually the youngest self-made billionaire in the world, which is pretty incredible. So he's a well-known personality in Silicon Valley, and he is arguably best known for how much he networks, which given that I do not live in San Francisco, Lauren and Mike can probably tell you a lot more about that. But he was once roommates with Sam Altman, who, supposedly, allegedly, told him to tone down the networking a notch. Unclear exactly what the motivations may have been there, but Wang clearly did not care for that advice and it has paid off for him. So this latest deal between Meta and Scale AI came actually after Wang and Zuckerberg reportedly spent a lot of one-on-one time together networking, presumably. And so it is really his relationships with so many key players in the AI industry that has positioned him now to be so powerful and to be in demand by a company like Meta that, as we'll talk about a little later, could really use a leg up in AI.
Bloomberg recently reported that Wang is known for calling "a dizzying amount of people," and not just senior, but junior staffers in AI firms, to know what they're working on and what they want to do. So he's not only a networker, but he's someone who keeps his ear to the ground. He's sort of in the know in Silicon Valley as far as AI goes.
Lauren Goode: It sounds exhausting.
Katie Drummond: It does. I felt tired just talking about that.
Lauren Goode: Yeah. Also, what does networking really look like in Silicon Valley? It's like, do you …
Katie Drummond: Well, you guys tell me. What do you do?
Lauren Goode: Yeah, I don't know. Meet some folks at Blue Bottle before you head on down to Marine Layer and then go to some all-night coding party in the carriage house of a Pacific Heights mansion? I don't know. What does that look like?
Michael Calore: It's a lot of walk and talks.
Katie Drummond: Is it really?
Lauren Goode: Well, in the Valley it is.
Michael Calore: Yeah.
Lauren Goode: Down around Palo Alto, people are in … Do people do that in San Francisco too?
Michael Calore: Oh, yeah.
Lauren Goode: Yeah? Yeah, that's a thing?
Michael Calore: Yeah, people do walk and talks.
Katie Drummond: Speaking of walking and talking, Mike, I want to hear more about the details of the deal.
Michael Calore: Yes. Let's talk about how much money is on the table on this deal, plus what Meta is hoping to get from the investment, and perhaps most importantly, what it is actually getting out of this deal.
Lauren Goode: So the deal was announced as a 14.3 billion dollar investment in Scale AI. To be clear here, this is not an acquisition. They're just taking a 49 percent stake in Scale AI and also bringing in Alexandr Wang and a bunch of talent, but it's not an acquisition, folks. Don't call it that. We've seen a wave of this over the past 12 to 18 months where Microsoft and Google have also either made strategic investments or licensing agreements with smaller AI players in order to sort of bring them into the fold, but not necessarily face the scrutiny of the U.S government and the Justice Department, because they want to move fast here. They want to build these AIs as quickly as possible, and so they want to just bring in all this talent. And so yes, we are lumping Meta in here, but that is essentially what they're doing with Scale AI.
Now for Alexandr, our master networker here, this deal also includes a leadership role in this superintelligence project that Meta is going to be building. We're going to get more into that later as we talk about superintelligence and what exactly it is. Zuckerberg, Meta now owns a strategic 49 percent stake in Scale AI. When it comes to what Meta is really getting from this deal, the main asset is the vast amount of AI training data that Scale AI possesses, and that leads to what Meta hopes to get to in the long run, a boost to the development of its Meta AI projects.
Katie Drummond: And I mean certainly Meta is doing a lot of clever things here, you could say, one of which is that they have put their competitors in a pretty tricky spot when it comes to whether or not they should continue working with Scale AI or whether they should move away to other companies. And Scale AI has other partnerships going on outside of the AI ecosystem as well. So the company has deals with foreign governments in Asia and Europe, it has a deal with the DOD actually for a first of its kind AI agent program, which sounds as dystopian as it probably is. It's called Thunder Forge. And the goal is essentially to enhance military operations with AI agents. I mean, it's exactly what you think it would be. And Meta could potentially benefit from all of those alliances that Scale AI has as well. So certainly you could say a mutually beneficial arrangement and one that puts a lot of players in the AI space on watch as it pertains to Meta.
Lauren Goode: I mean, I think the important thing to remember about Meta is that Meta buys its way into innovation. It doesn't spin it up entirely on its own. If you look at WhatsApp, if you look at Instagram, if you look at Oculus, all acquired by Meta. And then sometimes when Meta does try to spin up some vision for the future on its own, like the Metaverse, which is all we heard about two years ago, it falls flat on its face, or face computer, as we like to say. What Meta is doing here is it's locking up not only the data from Scale AI, but it's also locking up talent and technology that it believes is critical in sort of moving forward into this next phase of AI. An oversimplified way of looking at it would be, you need really two things to level up an AI right now, you need more compute power and you need a lot of data, and ideally high quality data.
This is clearly a data buy, and it's also a way of keeping other companies from potentially using it. Like Katie just said, Google is now backing out of its partnership with Scale AI.
Michael Calore: And Meta is doing this because it has a long way to go in order to reach the top of the heap, right? We've been talking about how Meta has been stumbling over the last few years in the AI race, so I want to dig into that a little bit, especially in how it has been lagging behind its competitors and what it's trying to do specifically to get ahead with this investment.
Lauren Goode: That's right, with this strategic investment, Mike. The short answer to that is Llama, which is its foundational model, its answer to OpenAI's ChatGPT and Google Gemini. Unfortunately so far, Llama hasn't really lived up to the hype around it. It's not as powerful as the rival systems of OpenAI and Google. Notably though Meta has tried to differentiate itself by open sourcing Llama, meaning that it freely shares the underlying code with outside software developers and businesses. So I guess it believes in the long run that that is actually going to be beneficial to it, even if right now it's not winning the speeds and feeds.
Michael Calore: Can I ask you, what's the prevailing wisdom on why Meta decided to release Llama as an open source model as opposed to keeping it closed like all of its competitors?
Lauren Goode: Yeah, it's a good question. I wish I could just call up Mark and ask him directly about this, but I can't, and we don't know exactly what's going on in his head. But I think in any robust technology environment, you're going to have lock-in and you're going to have open source providers, whether you're talking about IBM mainframes and then Linux, or you're talking about Apple software versus Google Android, and I think Google Android is the most used operating system in the world, and various products and business models and licensing models have extended from that. And so that may be part of what Mark Zuckerberg is thinking about long term. I think he also feels very burned by Apple. I mean, he has said that specifically when he wrote a blog post last year making a case for why Llama is open source. He said to do this well, we have to ensure that we always have access to the best technology and that we are not walking into a competitor's closed ecosystem where they can restrict what we build.
And he said that one of his most formative experiences, building Meta services, has been having those services constrained by what Apple will let Meta do on its platforms. So as he's thinking about the future of AI and what happens over the next several years or beyond that, I guess he doesn't want Meta to be locked in.
Michael Calore: Yeah, it feels like all of Meta's AI efforts have been very much on the surface, sort of lightweight consumer facing things, right? They have the Meta AI app that you can use to chat basically like a chatbot, they've incorporated their AI into the smart glasses and they've made these chatbots that live inside of Instagram and allow you to talk to somebody who talks like Snoop Dog basically. There's also the AI assisted search that has rolled out to all of the different Meta projects, but it really feels like this sort of social play in AI, specifically chatbot AI, has been the thing that they have just been concentrating on. They don't have advanced systems that corporations can license. They don't have a lot of the big ammunition that the other companies have.
Katie Drummond: Well, and the consumer facing AI, I mean if I may be so bold, is not particularly good or well-designed, or well-made, or a particularly good user experience. I mean, I don't have a Facebook account, and I haven't for a very long time. I do use Instagram. I mean, I have not once used that chatbot. So these consumer-facing bets have not been particularly successful in addition to not being very good, but they've also recently had some pretty significant blunders on the privacy front. Surprise, surprise, Meta having issues with privacy. Our colleague, Kylie Robison, reported that the app actually showed … This is insane. It is just insane. The app showed private conversations between the chatbot and users, including medical information, home addresses, even things directly related to pending court cases where people were talking to the chatbot, asking it questions, getting help with whatever weird sick question they were trying to answer or problem they were trying to solve.
And unbeknownst to them, those conversations were showing up in a social feed where everybody could see someone trying to break their tenant's lease or access some pornography or get a medical diagnosis for the weird lump on their foot. All of this stuff that, yeah, should you be careful in your conversations with any chatbot? Absolutely. But most people at least assume that those dialogues and those back and forths aren't just going to be published onto the internet. I mean, it's a pretty stunning failure. And it certainly doesn't feel accidental, it doesn't feel like they just were like, "Oh, how embarrassing that this has happened." I just don't think they really care. And this has been obviously the narrative around Meta for a very long time, which is, problems arise with their services, particularly as it pertains to privacy, and somebody out there in media reports on it, and then they patch it up, and that's exactly what has happened this time. I think they've put a disclaimer on top of the chatbot so that you know that you're opting in or opting out of some kind of public sharing of your conversations.
But it just sort of feels like they're not particularly interested in thinking very hard about the privacy piece, which, in my opinion, when it comes to chatbots and AI and sort of this brave new world we are all marching forward into or being marched forward into, in some cases, we should probably spend some time thinking about privacy considerations. Now, of course, with this deal with Scale AI, Meta is hoping to turn the page, to open a new chapter within all of their AI efforts, consumer facing or otherwise, specifically with this superintelligence AI lab.
Michael Calore: That's something that is really important that Meta would not be making all of these big statements about superintelligence. It would not be making this investment if it didn't feel like it had already failed in the race to develop strong AI programs. Because this is not just like a reorganization or realignment, this is a complete reset. It is a brand new team with new leadership and a new mission. For a very long time, the person who has been heading up Meta's AI efforts is Yann LeCun, who is a very well-respected Silicon Valley guru in AI. He has won the Turing Award, which is the nerd trophy for AI engineers. But LeCun famously does not think that artificial general intelligence is something that is on the immediate horizon. He's not a big chatbot proponent. He sees the value in large language models, but that's not where his interests are. And I think that if you're going to go full into, OK, what is the next thing? Then you need new blood. You need people who are true believers in this next phase, which, of course, is being defined by the term superintelligence.
Lauren Goode: Superintelligence.
Michael Calore: Let's take a break and then we'll come back and find out what that means.
Welcome back to Uncanny Valley . Before the break, we were mentioning that a key aspect of this deal between Meta and Scale AI, beyond all the big bucks involved, is the creation of a superintelligence AI lab. So question for the group, what the hell is superintelligence?
Katie Drummond: Lauren?
Lauren Goode: Oh, no. Well, it basically refers to developing an AI that goes beyond the human brain. It's a little bit unclear exactly what that means. For the past couple of years we've been hearing about AGI, which is artificial general intelligence, now we're talking about superintelligence. My understanding is, based on talking to researchers and technologists about this, is that it's not a flip the switch moment, it's not like there's going to be a specific model or some product release where all of a sudden we say, "Oh, we're living in AGI," And, "Oh, the next step is superintelligence." It's all sort of happening on a continuum. But the idea is that it is having something that is as smart as a human being or the human brain in your pocket on your phone, which is just crazy. It's not going to be sentient, it's not going to feel emotions, but it's going to do such a good job of replicating all of that that you're going to feel kind of blown away by it. And maybe that's bad news for us humans, I don't really know.
We do know that the term superintelligence was popularized by the Oxford philosopher, Nick Bostrom, who, in 2014, wrote a book on superintelligence. And he broke down a future where AI would advance to a point where it could turn against and delete humanity. And now, of course, it's being used by Mark Zuckerberg. So metaverse, superintelligence, I think we have a sense at this point of what's on Mark Zuckerberg's bookshelf.
Katie Drummond: And I mean, I think that whole question of sort of, what is AGI? What is superintelligence? I mean, so much of this is just branding. This is marketing that the AI industry is using to evoke a sense of, I think, sort of intimidation and fear and awe and respect and deference on the part of the general public, of policymakers, lawmakers. I mean, they want this thing to feel like this next era is right around the corner and we need to get ready now. And if we're not ready before China's ready, it's just going to be catastrophic because it's AGI. It's like, but what actually is that? And I think Meta is doing something super interesting and sort of cynical here, in my view, which is by positioning this new lab as superintelligence, they're essentially saying, "AGI? AGI is so last year. We're not even thinking about AGI. We're just jumping all the way to superintelligence."
I mean, this is marketing through and through. I'm not saying that this technology isn't evolving, that it won't drastically improve over time, that we won't see, and we already do see, AI that's capable of doing things that people can do. We see that all the time. There are plenty and plenty and plenty of examples of that. But I think that these terms are being used in a very squishy, opportunistic way for industry leaders and executives to sound and to make their technology sound as impressive and as valuable and as intimidating as possible. That's what I think.
Michael Calore: It is good marketing for recruiters too. If you're a person who's a professional in the AI industry, you're already pretty well paid, you feel like you're working on the next big thing because you're working on AGI somewhere, and then all of a sudden it's like, yeah, but wouldn't you rather be working on superintelligence?
Lauren Goode: Yeah. Yes.
Michael Calore: So who else is working on superintelligence? Are there other AGI companies that are like, "OK, no, wait, now we're doing that too." Are there people who have been working on superintelligence for a little while?
Lauren Goode: It depends on which AI visionary you're listening to. Sam Altman from OpenAI still seems pretty focused on AGI. He has said that he thinks it will be reached before the end of Trump's current presidential term. Dario Amadei from Anthropic has said that he thought AGI would happen in the next two years. So those are kind of still the AGI guys. On the other hand, you have Ilya Sutskever, who was the former chief scientist at OpenAI. He's cofounded a company called Safe Superintelligence, and they have a different approach. They are privately building superintelligence, and they say they will only release this technology to the world when it is deemed safe. So I think they're all kind of working towards the same thing, but superintelligence is the latest buzzword.
Katie Drummond: I love that they're going to be in hiding for many years.
Michael Calore: Yes.
Katie Drummond: I can't wait to see them crawl out of their bunker with their safe AI.
Michael Calore: Yeah, they're keeping it chained in the basement.
Katie Drummond: It is worth noting, Ilya aside, not all AI leaders are into the superintelligence hype. So these are people who have historically kept a lower profile, but more of them have actually started to speak up recently, which I think is notable. So Thomas Wolf is one example. He's Hugging Faces' cofounder and chief science officer. He called some parts of Amadei's Vision, "wishful thinking at best". That would be the vision of AGI. And Demis Hassabis, the CEO of Google DeepMind, has reportedly told his staff that in his opinion, the industry could be up to a decade away from developing AGI, noting that there is a lot that AI simply can't do today. So again, that's skepticism around AGI. We're not even talking about superintelligence. So God knows when that's going to be ready. I guess when Ilya lets us know.
Michael Calore: Let's talk about all of the people involved here. You really need all the top talent in order to really compete if you're going to build anything regardless of what you're calling it. And in this investment that Meta made with Scale AI, they get Alexandr Wang, comes to the company, he's bringing key people with him from Scale to work with him at Meta in the superintelligence lab. And this is happening at a time when AI talent is in super high demand with all the leading engineers being offered millions and tens of millions of dollars a year to work at the big companies. And apparently Meta has been offering up to nine figure compensation packages to get people to come work in the superintelligence lab.
Katie Drummond: Hold on. Can you just articulate nine figures? So that's not hundreds of thousands, it's not millions, it's not tens of millions, it's hundreds of millions?
Michael Calore: It is over $100 million.
Lauren Goode: What?
Michael Calore: Yes.
Katie Drummond: I mean, I'm speechless. I knew about the seven and eight figures, which is also just jaw dropping, but nine figures is unreal. That is unreal money.
Lauren Goode: I'm literally like, I got to read about this now. What? That is so crazy. I'm going to revise what I said earlier where I said, "Oh, the companies trying to level up in AI right now are looking at compute power and they're looking at data." They're also looking at talent. That is a huge, huge part of this. I'm still speechless that this is how much money these folks are getting offered, but I guess that's what Meta feels that it needs to do. I wonder how this is going to be reflected on its next earnings statement. We also saw that earlier this week, a longtime machine learning engineer and research scientist at OpenAI just got moved into a new position as the head of recruiting at OpenAI, which is fascinating. I mean, really, the initial reaction is kind of like, huh, that's an interesting career change. But then you think about, oh, this guy is going to talk the talk. He now has to go out and recruit top, top talent for OpenAI, continually recruit the top talent for OpenAI, and they now need to compete with Meta offering millions, bajillions of dollars to engineers.
So there is indeed, I think, a race for talent happening right now. Sequoia Capital investor, David Kahn, just wrote a blog post about this that I was reading where he did say that talent is the new bottleneck in AI, and he likened it to basically building a sports team. They're all backed by some mega rich tech company or individual, the star players can command these crazy pay packages in the tens of millions or hundreds of millions of dollars. Unlike sports teams though, where players often have long-term contracts, AI employment agreements can be short-term and liquid, which means that anyone can be poached at any time.
Katie Drummond: It's fascinating, it's grotesque. I want to know everything about how much these people are being paid. I want to know everything. And for Meta, it will be interesting to see how much money it takes for them to get top talent. I mean, money matters. If someone offered me a hundred million dollars to work for someone that I wasn't so excited about at a company that I thought had a so-so track record overall and a pretty poor one around AI, I mean, a hundred million dollars moves the needle. And it will be very interesting to see whether the Anthropics and OpenAIs and Googles of the world can compete. I mean, we know that Zuckerberg is doing a lot of this recruiting himself. He's personally reaching out to candidates. I wonder, and we will find out, whether that is to the company's benefit or not.
Lauren Goode: I mean, independent of how you might feel about Mark Zuckerberg too, Meta is just … It's been around for a while at this point. It's a twenty-year-old company. It's publicly traded. You'd probably get some really nice equity package on top of that too. But when you join a rocket ship like an OpenAI, you're joining because you think that if you get in early enough … At this point, it's not even that early, but that at some point you're going to become a multi, multi, multi-millionaire if that company either accelerates or it sells or something like that. And then that creates the flywheel effect that we always see in Silicon Valley, right? Early Google employees who left and went and started other things. We're going to see this wave eventually of OpenAI folks who leave and start other things. But if you're getting a comparable offer from OpenAI, or Meta at this point, which one are you going to go to when you're thinking about, really, the future?
Michael Calore: And I mean, there are a lot of people who won't take that money. I mean, if you think about the type of people who are commanding these super high salaries, they have been paid very, very well for a number of years, maybe their company was acquired and they had a big payout from that, so now they're sitting comfortably in a position at Anthropic or at OpenAI. And also the project that they're working on is something that matches their skills and maybe they want to see it through. So there's a bit of ego involved, there's a bit of life decision involved, and there's ethics involved, like, do you actually want to go work for Meta? Do you want to go build this thing that they're building? Especially after they made the announcement that they're going to start allowing their technologies to be used by the Department of Defense and they're going to start doing war stuff with the AI tools that they're building and the XR tools that they're building.
So yeah, I think there's a lot of people who are just sitting pretty right now and they have to decide whether or not that money is worth it to them as people.
Lauren Goode: Do you guys ever spend time on the app Blind?
Michael Calore: No. This is the one where people talk about what it's like to work at a company?
Lauren Goode: It is about working for a company because you have to affiliate yourself with a company when you sign up for the app, but it's all topics. And oftentimes it's people coming to the group with a compensation package and saying, I got offered this from Meta or Amazon and what should I do? And you just realize how distant this world is sometimes from the way the rest of the world or the rest of the country lives. People who are like, "Well, I don't know. Should I take this $700,000 package from Amazon? And that's without stock equity, benefits, or should I take this other comparable package from a similar company, but they're going to allow me to work from home? And I don't know though because I'm 40 and I only have 10 million in retirement." I'm like, oh my God, that is the world we're talking about.
Michael Calore: Yeah.
Katie Drummond: We need to do more reporting on this. I think that the compensation of people in Silicon Valley is fascinating.
Lauren Goode: Well, if anyone would like to weigh in, if you're a recruiter, if you're a person who's been made one of these offers from the Meta superintelligence lab, we want to hear from you.
Michael Calore: Big money.
Katie Drummond: We sure do.
Lauren Goode: Our signals are out there.
Michael Calore: Big money, no whammies.
Lauren Goode: Now we know what Katie would leave us for to go work for Mark Zuckerberg.
Katie Drummond: A hundred million dollars is a lot of money. It's a lot of money.
Lauren Goode: It's a lot of money.
Katie Drummond: That would be tough for me. I don't think I could do it.
Lauren Goode: Yep. If you invest it, well, it'd be a lot of money for your kids' kids' kids.
Katie Drummond: I know, but then I'd have to tell my kid what I do, and I don't know that I could do that. I'm being totally honest. I don't think I could do it. Let me be clear, there are a lot of fantastic people who work at Meta. I mean, this is not a repudiation of anyone's decisions or career choices or where they have chosen to work, given my background and what I do for a living, yeah, I don't know. I don't think I could do that.
Michael Calore: You get to be part of the superintelligence revolution.
Katie Drummond: I don't want to.
Michael Calore: Maybe just use the chatbot and then you can feel like you're a part of it.
Katie Drummond: Yeah, there you go. I have some pressing and highly personal questions for Meta's chatbot, and as soon as we get off this recording, I'm going to go ask all of them in private.
Michael Calore: I look forward to reading them on the [inaudible 00:30:11].
Lauren Goode: Katie's like, how do I extract myself from a work project that has me locked in a room for two hours every week?
Katie Drummond: Oh dear.
Michael Calore: OK, let's take another break and we'll come right back with recommendations.
All right, thank you both for a great conversation about superintelligence. So I think it's time to give our listeners something from our own superintelligent human brains, our recommendations for the week. Lauren, would you like to go first?
Lauren Goode: Sure. I recently learned that by using generative AI tools like ChatGPT, you can get your color analysis done. Have either of you ever done this?
Michael Calore: No.
Katie Drummond: No.
Lauren Goode: So this is a thing that is part of the beauty influencer world online where typically you would pay someone, sometimes a human, sometimes an app that has human input, to analyze the color of your hair, skin, eyes, skin tone, all that, and tell you what season you are and then tell you what clothing you should wear in a way that accentuates your whole situation.
Katie Drummond: Did you find this revelatory?
Lauren Goode: Yeah. So recently when I was hanging out with some friends, one of them had had her color analysis done and we were talking about it and she said, "Oh, you can just do it on ChatGPT." And I was like, "What? You don't have to pay someone a couple hundred dollars to do this?" So she uploaded our photos into ChatGPT and I got a color analysis done. And so I'm a deep autumn, in case anyone wants to know.
Michael Calore: It's a burning question.
Lauren Goode: That was the burning question. I thought maybe I was a winter, but I'm a deep autumn. And so I think to date, this is the most interesting use case of ChatGPT I've experienced so far.
Michael Calore: Ever, of all …
Lauren Goode: No, that's not true.
Michael Calore: Of all your experiences.
Lauren Goode: That's not true. The other day it told me how to cook salini mushrooms instead of cremini mushrooms. I'm quite certain salini mushrooms don't exist. So it was helpful in that regard too.
Katie Drummond: Yikes.
Lauren Goode: But no, I've used it for other things too. I've used it for research and reasoning and fun things like that.
Michael Calore: That's pretty good.
Lauren Goode: So maybe later, actually, I'm going to upload photos of you both to ChatGPT and ask it to do your color analysis.
Michael Calore: I do not consent.
Lauren Goode: And then I'll tell you … OK. What color you should be wearing. Both of you are wearing all black right now.
Katie Drummond: As usual.
Michael Calore: Yeah, I was going to say. I call those days weekdays.
Lauren Goode: Anarchists in the room. I love this. OK, that's it. That's my recommendation.
Katie Drummond: I was thinking about what to recommend and it was either going to be a book, a kid's movie or food, and I was like, no, girl, you've done all of those already. You have to think of something else. I have an AI-related recommendation actually, which is, I had someone come recently to help me out with my outdoor plants, like our little backyard area in Brooklyn, and I was telling him that all of my indoor plants are dying and struggling, and I was just so confused. I was like, which window should this one go in? And what … Blah, blah, blah, blah, blah. How often do I water all of these things? I have many plants. And he recommended this app called Picture This. And you take photos of all your plants and you upload the photos into the app, it tells you what kind of plant it is, it tells you how often to water it. You can use your phone to show the app how much sun is coming in through the window, and it'll tell you if that's enough sun, too much sun, not enough sun.
Oh, and you can take a photo of the plant and it'll tell you if it's sick and what it is struggling with, which is very upsetting, but very helpful. I have many struggling plants. But it's very, very cool. It's definitely highly judgmental. I get push notifications now saying, "Don't you want to take care of your plants?" And I'm like, well, I do, but I also have a job. It's a very, very interesting product, and if you have plants, I highly recommend it.
Lauren Goode: Picture this.
Katie Drummond: Picture this. Healthy plants, healthy you. Superintelligence, picture this.
Lauren Goode: Mike, what's your recommendation?
Michael Calore: I think given the current geopolitical state of the world, it is time to re-watch Dr. Strangelove , the 1964 film by Stanley Kubrick. Dr. Strangelove or: How I Learned to Stop Worrying and Love the Bomb . It's about escalating political tensions based on a misunderstanding that leads to nuclear war. It's a farce. It's very funny. It's also very dry, but it's a great movie. I re-watch it about once a year, unfortunately. I'm usually compelled by current events to watch it. So I would say it is a good time to watch Dr. Strangelove . And I'm not saying the world is going to end, but whenever we start talking about nuclear power and we start talking about the people in the world who have their fingers on the button, it is important to remind ourselves that these are human decisions that people make about our future, and it's a great movie to help you process that information.
Katie Drummond: Wow.
Lauren Goode: So Katie's watering her plants and I'm cutting bangs and you are watching movies about the end of the world. I see who among us is really diving in and who's …
Michael Calore: So a lot of people ignore the void. A lot of people acknowledge it. I'm one of those people who puts my face right up against the glass and just scares at it.
Katie Drummond: That's just like my husband. Not to make it all about him, but … I feel like I cover the void. I do the void for my job. Every day, I'm in the void. So when I'm out of the work void, I want to go into la la land. And so last night I was telling people in Slack today, my husband was texting me links to used motorboats because he was like the safest place for us to be during a nuclear attack is in the middle of a body of water.
Lauren Goode: Good lord.
Katie Drummond: And I was like, and what exactly do you think we're going to be doing in this little motorboat in the middle of the Atlantic Ocean off the coast of New York City? I think if it comes down to motorboat or death, we are probably going to die.
Lauren Goode: Yeah, yeah.
Katie Drummond: Just saying.
Lauren Goode: That was before bed, like before you were supposed to go to sleep?
Katie Drummond: Right as I was lying in bed trying to go to sleep and the thing is like ping, ping, links to used boats.
Lauren Goode: I mean, I think we're all sort of void adjacent these days, so you can't really ignore it.
Katie Drummond: You can't ignore it, but it's a choice to lean into it in your personal time, I will say.
Michael Calore: I would say that if you're going to do that exercise, there are fewer people who it's more delightful to do it with than Peter Sellers and George C. Scott and Sterling Hayden and Stanley Kubrick. It's a great movie, so stream it tonight.
Katie Drummond: We'll link to it in the show notes.
Michael Calore: Thanks for listening to Uncanny Valley . If you liked what you heard today, make sure to follow our show and rate it on your podcast app of choice. If you'd like to get in touch with us with any questions, comments, or show suggestions, write to us at uncannyvalley@wired.com. Today's show is produced by Kyana Moghadam and Adriana Tapia. Amar Lal at Macrosound mixed this episode. Jake Lummus was our New York studio engineer. Daniel Roman fact-checked this episode. Jordan Bell is our executive producer. Katie Drummond is WIRED's global editorial director, and Chris Bannon is the head of global audio.
Today on the show, we dive into the deal between Meta and Scale AI, including what Meta aims to get out of investment, and we ask the question we are all wondering: What is superhuman intelligence, anyway?
You can follow Michael Calore on Bluesky at @snackfight, Lauren Goode on Bluesky at @laurengoode, and Katie Drummond on Bluesky at @katie-drummond. Write to us at uncannyvalley@wired.com. How to Listen
You can always listen to this week's podcast through the audio player on this page, but if you want to subscribe for free to get every episode, here's how:
If you're on an iPhone or iPad, open the app called Podcasts, or just tap this link. You can also download an app like Overcast or Pocket Casts and search for 'uncanny valley.' We're on Spotify too. Transcript
Note: This is an automated transcript, which may contain errors.
Michael Calore: How is everybody doing this week?
Katie Drummond: Well, I'm back and I'm so happy to be here. I ate so much butter in France last week, the first couple nights I had to myself. And so what would a normal person do in Paris? Maybe they'd go out and sit at the bar and have dinner alone, maybe they'd meet up with a friend. I ate butter and bread alone in my hotel room. And let me tell you, if you're listening out there and you are a mom and you have a young kid or young kids, if you're married, if you struggle with having a spouse and a child and maybe some pets and a busy job, there is no better experience than eating French butter and bread alone in a hotel room.
Michael Calore: Wow.
Lauren Goode: Uncanny Valley , brought to you by the dairy lobby.
Katie Drummond: By the French dairy lobby. I feel incredible. How are you guys?
Lauren Goode: I'm OK. I got bangs.
Katie Drummond: You did?
Lauren Goode: Yeah. So most of our listeners can't see it unless you watch our new video promos online, but I got bangs. There's often a correlation between things going on in the world and women cutting their bangs. That's all I'm going to say about that. But otherwise, I'm great. I'm great. Rate the bangs, go online. Thumbs up, thumbs down.
Katie Drummond: Five out of five.
Lauren Goode: Thank you. Katie, really, you were my inspiration here.
Katie Drummond: Oh, no, that's too kind. But I do love a bang and hate a forehead.
Michael Calore: Well, have I got a haircut for you.
Lauren Goode: Well, Mike, how are you doing? Sorry about that, Mike.
Katie Drummond: But there's just so much smarts in there, Mike. That's the thing.
Michael Calore: I'm just as God made me.
Katie Drummond: How are you doing?
Michael Calore: I'm doing great. I don't have any hair stories. I haven't eaten any butter recently, so I'm feeling really left out right now. This is WIRED's Uncanny Valley , a show about the people, power and influence of Silicon Valley. Today we're talking about Meta's recent investment in Scale AI, and its move to build a superintelligence AI research lab. It's the latest effort from Meta to compete with the big names in the AI race like OpenAI, Anthropic, and Google. But Meta is taking a different approach. Not only are its AI models open source, but in typical Meta fashion, it seems to be trying to outspend its competitors to acquire top talent, and its Scale AI investment, which is not an acquisition, is just part of that strategy. We'll dive into what Meta is hoping to get from this investment and what it's actually getting, and whether this move could give the company the competitive advantage it's seeking. Plus we will ask, what is superintelligence anyway? I'm Michael Calore, director of consumer tech and culture here at WIRED.
Lauren Goode: I'm Lauren Goode. I'm a senior correspondent at WIRED.
Katie Drummond: And I'm Katie Drummond, WIRED's global editorial director.
Michael Calore: So let's start off by diving into Scale AI. Unlike Meta, the company is not what you would call a household name anywhere outside of Silicon Valley, but it's certainly made a name for itself in the AI world. What's up with Scale?
Lauren Goode: Scale AI is a data labeling company. Sounds very sexy, doesn't it? They do the grunt work of analyzing and categorizing the data that is later distributed to larger AI models in order to train them. It's nothing fancy, but they kind of perform an essential function for machine learning programs to improve. And they have some pretty big customers. Companies like OpenAI and Google have been among their clients. Earlier this year, our colleague Will Knight reported on Scale AI's new platform that allows AI models to be automatically tested against key benchmarks and pinpoint any weaknesses in the models. So basically Scale AI has been making this concentrated effort to be a key partner as it's working with these larger AI companies.
Katie Drummond: I think it's also worth pointing out that really at the heart of the company's success is its founder, Alexandr Wang, without an E, right?
Michael Calore: Yeah.
Lauren Goode: Yeah.
Michael Calore: He's the Flickr of CEOs.
Katie Drummond: There you go. He's 28 years old, and at one point in time he was actually the youngest self-made billionaire in the world, which is pretty incredible. So he's a well-known personality in Silicon Valley, and he is arguably best known for how much he networks, which given that I do not live in San Francisco, Lauren and Mike can probably tell you a lot more about that. But he was once roommates with Sam Altman, who, supposedly, allegedly, told him to tone down the networking a notch. Unclear exactly what the motivations may have been there, but Wang clearly did not care for that advice and it has paid off for him. So this latest deal between Meta and Scale AI came actually after Wang and Zuckerberg reportedly spent a lot of one-on-one time together networking, presumably. And so it is really his relationships with so many key players in the AI industry that has positioned him now to be so powerful and to be in demand by a company like Meta that, as we'll talk about a little later, could really use a leg up in AI.
Bloomberg recently reported that Wang is known for calling "a dizzying amount of people," and not just senior, but junior staffers in AI firms, to know what they're working on and what they want to do. So he's not only a networker, but he's someone who keeps his ear to the ground. He's sort of in the know in Silicon Valley as far as AI goes.
Lauren Goode: It sounds exhausting.
Katie Drummond: It does. I felt tired just talking about that.
Lauren Goode: Yeah. Also, what does networking really look like in Silicon Valley? It's like, do you …
Katie Drummond: Well, you guys tell me. What do you do?
Lauren Goode: Yeah, I don't know. Meet some folks at Blue Bottle before you head on down to Marine Layer and then go to some all-night coding party in the carriage house of a Pacific Heights mansion? I don't know. What does that look like?
Michael Calore: It's a lot of walk and talks.
Katie Drummond: Is it really?
Lauren Goode: Well, in the Valley it is.
Michael Calore: Yeah.
Lauren Goode: Down around Palo Alto, people are in … Do people do that in San Francisco too?
Michael Calore: Oh, yeah.
Lauren Goode: Yeah? Yeah, that's a thing?
Michael Calore: Yeah, people do walk and talks.
Katie Drummond: Speaking of walking and talking, Mike, I want to hear more about the details of the deal.
Michael Calore: Yes. Let's talk about how much money is on the table on this deal, plus what Meta is hoping to get from the investment, and perhaps most importantly, what it is actually getting out of this deal.
Lauren Goode: So the deal was announced as a 14.3 billion dollar investment in Scale AI. To be clear here, this is not an acquisition. They're just taking a 49 percent stake in Scale AI and also bringing in Alexandr Wang and a bunch of talent, but it's not an acquisition, folks. Don't call it that. We've seen a wave of this over the past 12 to 18 months where Microsoft and Google have also either made strategic investments or licensing agreements with smaller AI players in order to sort of bring them into the fold, but not necessarily face the scrutiny of the U.S government and the Justice Department, because they want to move fast here. They want to build these AIs as quickly as possible, and so they want to just bring in all this talent. And so yes, we are lumping Meta in here, but that is essentially what they're doing with Scale AI.
Now for Alexandr, our master networker here, this deal also includes a leadership role in this superintelligence project that Meta is going to be building. We're going to get more into that later as we talk about superintelligence and what exactly it is. Zuckerberg, Meta now owns a strategic 49 percent stake in Scale AI. When it comes to what Meta is really getting from this deal, the main asset is the vast amount of AI training data that Scale AI possesses, and that leads to what Meta hopes to get to in the long run, a boost to the development of its Meta AI projects.
Katie Drummond: And I mean certainly Meta is doing a lot of clever things here, you could say, one of which is that they have put their competitors in a pretty tricky spot when it comes to whether or not they should continue working with Scale AI or whether they should move away to other companies. And Scale AI has other partnerships going on outside of the AI ecosystem as well. So the company has deals with foreign governments in Asia and Europe, it has a deal with the DOD actually for a first of its kind AI agent program, which sounds as dystopian as it probably is. It's called Thunder Forge. And the goal is essentially to enhance military operations with AI agents. I mean, it's exactly what you think it would be. And Meta could potentially benefit from all of those alliances that Scale AI has as well. So certainly you could say a mutually beneficial arrangement and one that puts a lot of players in the AI space on watch as it pertains to Meta.
Lauren Goode: I mean, I think the important thing to remember about Meta is that Meta buys its way into innovation. It doesn't spin it up entirely on its own. If you look at WhatsApp, if you look at Instagram, if you look at Oculus, all acquired by Meta. And then sometimes when Meta does try to spin up some vision for the future on its own, like the Metaverse, which is all we heard about two years ago, it falls flat on its face, or face computer, as we like to say. What Meta is doing here is it's locking up not only the data from Scale AI, but it's also locking up talent and technology that it believes is critical in sort of moving forward into this next phase of AI. An oversimplified way of looking at it would be, you need really two things to level up an AI right now, you need more compute power and you need a lot of data, and ideally high quality data.
This is clearly a data buy, and it's also a way of keeping other companies from potentially using it. Like Katie just said, Google is now backing out of its partnership with Scale AI.
Michael Calore: And Meta is doing this because it has a long way to go in order to reach the top of the heap, right? We've been talking about how Meta has been stumbling over the last few years in the AI race, so I want to dig into that a little bit, especially in how it has been lagging behind its competitors and what it's trying to do specifically to get ahead with this investment.
Lauren Goode: That's right, with this strategic investment, Mike. The short answer to that is Llama, which is its foundational model, its answer to OpenAI's ChatGPT and Google Gemini. Unfortunately so far, Llama hasn't really lived up to the hype around it. It's not as powerful as the rival systems of OpenAI and Google. Notably though Meta has tried to differentiate itself by open sourcing Llama, meaning that it freely shares the underlying code with outside software developers and businesses. So I guess it believes in the long run that that is actually going to be beneficial to it, even if right now it's not winning the speeds and feeds.
Michael Calore: Can I ask you, what's the prevailing wisdom on why Meta decided to release Llama as an open source model as opposed to keeping it closed like all of its competitors?
Lauren Goode: Yeah, it's a good question. I wish I could just call up Mark and ask him directly about this, but I can't, and we don't know exactly what's going on in his head. But I think in any robust technology environment, you're going to have lock-in and you're going to have open source providers, whether you're talking about IBM mainframes and then Linux, or you're talking about Apple software versus Google Android, and I think Google Android is the most used operating system in the world, and various products and business models and licensing models have extended from that. And so that may be part of what Mark Zuckerberg is thinking about long term. I think he also feels very burned by Apple. I mean, he has said that specifically when he wrote a blog post last year making a case for why Llama is open source. He said to do this well, we have to ensure that we always have access to the best technology and that we are not walking into a competitor's closed ecosystem where they can restrict what we build.
And he said that one of his most formative experiences, building Meta services, has been having those services constrained by what Apple will let Meta do on its platforms. So as he's thinking about the future of AI and what happens over the next several years or beyond that, I guess he doesn't want Meta to be locked in.
Michael Calore: Yeah, it feels like all of Meta's AI efforts have been very much on the surface, sort of lightweight consumer facing things, right? They have the Meta AI app that you can use to chat basically like a chatbot, they've incorporated their AI into the smart glasses and they've made these chatbots that live inside of Instagram and allow you to talk to somebody who talks like Snoop Dog basically. There's also the AI assisted search that has rolled out to all of the different Meta projects, but it really feels like this sort of social play in AI, specifically chatbot AI, has been the thing that they have just been concentrating on. They don't have advanced systems that corporations can license. They don't have a lot of the big ammunition that the other companies have.
Katie Drummond: Well, and the consumer facing AI, I mean if I may be so bold, is not particularly good or well-designed, or well-made, or a particularly good user experience. I mean, I don't have a Facebook account, and I haven't for a very long time. I do use Instagram. I mean, I have not once used that chatbot. So these consumer-facing bets have not been particularly successful in addition to not being very good, but they've also recently had some pretty significant blunders on the privacy front. Surprise, surprise, Meta having issues with privacy. Our colleague, Kylie Robison, reported that the app actually showed … This is insane. It is just insane. The app showed private conversations between the chatbot and users, including medical information, home addresses, even things directly related to pending court cases where people were talking to the chatbot, asking it questions, getting help with whatever weird sick question they were trying to answer or problem they were trying to solve.
And unbeknownst to them, those conversations were showing up in a social feed where everybody could see someone trying to break their tenant's lease or access some pornography or get a medical diagnosis for the weird lump on their foot. All of this stuff that, yeah, should you be careful in your conversations with any chatbot? Absolutely. But most people at least assume that those dialogues and those back and forths aren't just going to be published onto the internet. I mean, it's a pretty stunning failure. And it certainly doesn't feel accidental, it doesn't feel like they just were like, "Oh, how embarrassing that this has happened." I just don't think they really care. And this has been obviously the narrative around Meta for a very long time, which is, problems arise with their services, particularly as it pertains to privacy, and somebody out there in media reports on it, and then they patch it up, and that's exactly what has happened this time. I think they've put a disclaimer on top of the chatbot so that you know that you're opting in or opting out of some kind of public sharing of your conversations.
But it just sort of feels like they're not particularly interested in thinking very hard about the privacy piece, which, in my opinion, when it comes to chatbots and AI and sort of this brave new world we are all marching forward into or being marched forward into, in some cases, we should probably spend some time thinking about privacy considerations. Now, of course, with this deal with Scale AI, Meta is hoping to turn the page, to open a new chapter within all of their AI efforts, consumer facing or otherwise, specifically with this superintelligence AI lab.
Michael Calore: That's something that is really important that Meta would not be making all of these big statements about superintelligence. It would not be making this investment if it didn't feel like it had already failed in the race to develop strong AI programs. Because this is not just like a reorganization or realignment, this is a complete reset. It is a brand new team with new leadership and a new mission. For a very long time, the person who has been heading up Meta's AI efforts is Yann LeCun, who is a very well-respected Silicon Valley guru in AI. He has won the Turing Award, which is the nerd trophy for AI engineers. But LeCun famously does not think that artificial general intelligence is something that is on the immediate horizon. He's not a big chatbot proponent. He sees the value in large language models, but that's not where his interests are. And I think that if you're going to go full into, OK, what is the next thing? Then you need new blood. You need people who are true believers in this next phase, which, of course, is being defined by the term superintelligence.
Lauren Goode: Superintelligence.
Michael Calore: Let's take a break and then we'll come back and find out what that means.
Welcome back to Uncanny Valley . Before the break, we were mentioning that a key aspect of this deal between Meta and Scale AI, beyond all the big bucks involved, is the creation of a superintelligence AI lab. So question for the group, what the hell is superintelligence?
Katie Drummond: Lauren?
Lauren Goode: Oh, no. Well, it basically refers to developing an AI that goes beyond the human brain. It's a little bit unclear exactly what that means. For the past couple of years we've been hearing about AGI, which is artificial general intelligence, now we're talking about superintelligence. My understanding is, based on talking to researchers and technologists about this, is that it's not a flip the switch moment, it's not like there's going to be a specific model or some product release where all of a sudden we say, "Oh, we're living in AGI," And, "Oh, the next step is superintelligence." It's all sort of happening on a continuum. But the idea is that it is having something that is as smart as a human being or the human brain in your pocket on your phone, which is just crazy. It's not going to be sentient, it's not going to feel emotions, but it's going to do such a good job of replicating all of that that you're going to feel kind of blown away by it. And maybe that's bad news for us humans, I don't really know.
We do know that the term superintelligence was popularized by the Oxford philosopher, Nick Bostrom, who, in 2014, wrote a book on superintelligence. And he broke down a future where AI would advance to a point where it could turn against and delete humanity. And now, of course, it's being used by Mark Zuckerberg. So metaverse, superintelligence, I think we have a sense at this point of what's on Mark Zuckerberg's bookshelf.
Katie Drummond: And I mean, I think that whole question of sort of, what is AGI? What is superintelligence? I mean, so much of this is just branding. This is marketing that the AI industry is using to evoke a sense of, I think, sort of intimidation and fear and awe and respect and deference on the part of the general public, of policymakers, lawmakers. I mean, they want this thing to feel like this next era is right around the corner and we need to get ready now. And if we're not ready before China's ready, it's just going to be catastrophic because it's AGI. It's like, but what actually is that? And I think Meta is doing something super interesting and sort of cynical here, in my view, which is by positioning this new lab as superintelligence, they're essentially saying, "AGI? AGI is so last year. We're not even thinking about AGI. We're just jumping all the way to superintelligence."
I mean, this is marketing through and through. I'm not saying that this technology isn't evolving, that it won't drastically improve over time, that we won't see, and we already do see, AI that's capable of doing things that people can do. We see that all the time. There are plenty and plenty and plenty of examples of that. But I think that these terms are being used in a very squishy, opportunistic way for industry leaders and executives to sound and to make their technology sound as impressive and as valuable and as intimidating as possible. That's what I think.
Michael Calore: It is good marketing for recruiters too. If you're a person who's a professional in the AI industry, you're already pretty well paid, you feel like you're working on the next big thing because you're working on AGI somewhere, and then all of a sudden it's like, yeah, but wouldn't you rather be working on superintelligence?
Lauren Goode: Yeah. Yes.
Michael Calore: So who else is working on superintelligence? Are there other AGI companies that are like, "OK, no, wait, now we're doing that too." Are there people who have been working on superintelligence for a little while?
Lauren Goode: It depends on which AI visionary you're listening to. Sam Altman from OpenAI still seems pretty focused on AGI. He has said that he thinks it will be reached before the end of Trump's current presidential term. Dario Amadei from Anthropic has said that he thought AGI would happen in the next two years. So those are kind of still the AGI guys. On the other hand, you have Ilya Sutskever, who was the former chief scientist at OpenAI. He's cofounded a company called Safe Superintelligence, and they have a different approach. They are privately building superintelligence, and they say they will only release this technology to the world when it is deemed safe. So I think they're all kind of working towards the same thing, but superintelligence is the latest buzzword.
Katie Drummond: I love that they're going to be in hiding for many years.
Michael Calore: Yes.
Katie Drummond: I can't wait to see them crawl out of their bunker with their safe AI.
Michael Calore: Yeah, they're keeping it chained in the basement.
Katie Drummond: It is worth noting, Ilya aside, not all AI leaders are into the superintelligence hype. So these are people who have historically kept a lower profile, but more of them have actually started to speak up recently, which I think is notable. So Thomas Wolf is one example. He's Hugging Faces' cofounder and chief science officer. He called some parts of Amadei's Vision, "wishful thinking at best". That would be the vision of AGI. And Demis Hassabis, the CEO of Google DeepMind, has reportedly told his staff that in his opinion, the industry could be up to a decade away from developing AGI, noting that there is a lot that AI simply can't do today. So again, that's skepticism around AGI. We're not even talking about superintelligence. So God knows when that's going to be ready. I guess when Ilya lets us know.
Michael Calore: Let's talk about all of the people involved here. You really need all the top talent in order to really compete if you're going to build anything regardless of what you're calling it. And in this investment that Meta made with Scale AI, they get Alexandr Wang, comes to the company, he's bringing key people with him from Scale to work with him at Meta in the superintelligence lab. And this is happening at a time when AI talent is in super high demand with all the leading engineers being offered millions and tens of millions of dollars a year to work at the big companies. And apparently Meta has been offering up to nine figure compensation packages to get people to come work in the superintelligence lab.
Katie Drummond: Hold on. Can you just articulate nine figures? So that's not hundreds of thousands, it's not millions, it's not tens of millions, it's hundreds of millions?
Michael Calore: It is over $100 million.
Lauren Goode: What?
Michael Calore: Yes.
Katie Drummond: I mean, I'm speechless. I knew about the seven and eight figures, which is also just jaw dropping, but nine figures is unreal. That is unreal money.
Lauren Goode: I'm literally like, I got to read about this now. What? That is so crazy. I'm going to revise what I said earlier where I said, "Oh, the companies trying to level up in AI right now are looking at compute power and they're looking at data." They're also looking at talent. That is a huge, huge part of this. I'm still speechless that this is how much money these folks are getting offered, but I guess that's what Meta feels that it needs to do. I wonder how this is going to be reflected on its next earnings statement. We also saw that earlier this week, a longtime machine learning engineer and research scientist at OpenAI just got moved into a new position as the head of recruiting at OpenAI, which is fascinating. I mean, really, the initial reaction is kind of like, huh, that's an interesting career change. But then you think about, oh, this guy is going to talk the talk. He now has to go out and recruit top, top talent for OpenAI, continually recruit the top talent for OpenAI, and they now need to compete with Meta offering millions, bajillions of dollars to engineers.
So there is indeed, I think, a race for talent happening right now. Sequoia Capital investor, David Kahn, just wrote a blog post about this that I was reading where he did say that talent is the new bottleneck in AI, and he likened it to basically building a sports team. They're all backed by some mega rich tech company or individual, the star players can command these crazy pay packages in the tens of millions or hundreds of millions of dollars. Unlike sports teams though, where players often have long-term contracts, AI employment agreements can be short-term and liquid, which means that anyone can be poached at any time.
Katie Drummond: It's fascinating, it's grotesque. I want to know everything about how much these people are being paid. I want to know everything. And for Meta, it will be interesting to see how much money it takes for them to get top talent. I mean, money matters. If someone offered me a hundred million dollars to work for someone that I wasn't so excited about at a company that I thought had a so-so track record overall and a pretty poor one around AI, I mean, a hundred million dollars moves the needle. And it will be very interesting to see whether the Anthropics and OpenAIs and Googles of the world can compete. I mean, we know that Zuckerberg is doing a lot of this recruiting himself. He's personally reaching out to candidates. I wonder, and we will find out, whether that is to the company's benefit or not.
Lauren Goode: I mean, independent of how you might feel about Mark Zuckerberg too, Meta is just … It's been around for a while at this point. It's a twenty-year-old company. It's publicly traded. You'd probably get some really nice equity package on top of that too. But when you join a rocket ship like an OpenAI, you're joining because you think that if you get in early enough … At this point, it's not even that early, but that at some point you're going to become a multi, multi, multi-millionaire if that company either accelerates or it sells or something like that. And then that creates the flywheel effect that we always see in Silicon Valley, right? Early Google employees who left and went and started other things. We're going to see this wave eventually of OpenAI folks who leave and start other things. But if you're getting a comparable offer from OpenAI, or Meta at this point, which one are you going to go to when you're thinking about, really, the future?
Michael Calore: And I mean, there are a lot of people who won't take that money. I mean, if you think about the type of people who are commanding these super high salaries, they have been paid very, very well for a number of years, maybe their company was acquired and they had a big payout from that, so now they're sitting comfortably in a position at Anthropic or at OpenAI. And also the project that they're working on is something that matches their skills and maybe they want to see it through. So there's a bit of ego involved, there's a bit of life decision involved, and there's ethics involved, like, do you actually want to go work for Meta? Do you want to go build this thing that they're building? Especially after they made the announcement that they're going to start allowing their technologies to be used by the Department of Defense and they're going to start doing war stuff with the AI tools that they're building and the XR tools that they're building.
So yeah, I think there's a lot of people who are just sitting pretty right now and they have to decide whether or not that money is worth it to them as people.
Lauren Goode: Do you guys ever spend time on the app Blind?
Michael Calore: No. This is the one where people talk about what it's like to work at a company?
Lauren Goode: It is about working for a company because you have to affiliate yourself with a company when you sign up for the app, but it's all topics. And oftentimes it's people coming to the group with a compensation package and saying, I got offered this from Meta or Amazon and what should I do? And you just realize how distant this world is sometimes from the way the rest of the world or the rest of the country lives. People who are like, "Well, I don't know. Should I take this $700,000 package from Amazon? And that's without stock equity, benefits, or should I take this other comparable package from a similar company, but they're going to allow me to work from home? And I don't know though because I'm 40 and I only have 10 million in retirement." I'm like, oh my God, that is the world we're talking about.
Michael Calore: Yeah.
Katie Drummond: We need to do more reporting on this. I think that the compensation of people in Silicon Valley is fascinating.
Lauren Goode: Well, if anyone would like to weigh in, if you're a recruiter, if you're a person who's been made one of these offers from the Meta superintelligence lab, we want to hear from you.
Michael Calore: Big money.
Katie Drummond: We sure do.
Lauren Goode: Our signals are out there.
Michael Calore: Big money, no whammies.
Lauren Goode: Now we know what Katie would leave us for to go work for Mark Zuckerberg.
Katie Drummond: A hundred million dollars is a lot of money. It's a lot of money.
Lauren Goode: It's a lot of money.
Katie Drummond: That would be tough for me. I don't think I could do it.
Lauren Goode: Yep. If you invest it, well, it'd be a lot of money for your kids' kids' kids.
Katie Drummond: I know, but then I'd have to tell my kid what I do, and I don't know that I could do that. I'm being totally honest. I don't think I could do it. Let me be clear, there are a lot of fantastic people who work at Meta. I mean, this is not a repudiation of anyone's decisions or career choices or where they have chosen to work, given my background and what I do for a living, yeah, I don't know. I don't think I could do that.
Michael Calore: You get to be part of the superintelligence revolution.
Katie Drummond: I don't want to.
Michael Calore: Maybe just use the chatbot and then you can feel like you're a part of it.
Katie Drummond: Yeah, there you go. I have some pressing and highly personal questions for Meta's chatbot, and as soon as we get off this recording, I'm going to go ask all of them in private.
Michael Calore: I look forward to reading them on the [inaudible 00:30:11].
Lauren Goode: Katie's like, how do I extract myself from a work project that has me locked in a room for two hours every week?
Katie Drummond: Oh dear.
Michael Calore: OK, let's take another break and we'll come right back with recommendations.
All right, thank you both for a great conversation about superintelligence. So I think it's time to give our listeners something from our own superintelligent human brains, our recommendations for the week. Lauren, would you like to go first?
Lauren Goode: Sure. I recently learned that by using generative AI tools like ChatGPT, you can get your color analysis done. Have either of you ever done this?
Michael Calore: No.
Katie Drummond: No.
Lauren Goode: So this is a thing that is part of the beauty influencer world online where typically you would pay someone, sometimes a human, sometimes an app that has human input, to analyze the color of your hair, skin, eyes, skin tone, all that, and tell you what season you are and then tell you what clothing you should wear in a way that accentuates your whole situation.
Katie Drummond: Did you find this revelatory?
Lauren Goode: Yeah. So recently when I was hanging out with some friends, one of them had had her color analysis done and we were talking about it and she said, "Oh, you can just do it on ChatGPT." And I was like, "What? You don't have to pay someone a couple hundred dollars to do this?" So she uploaded our photos into ChatGPT and I got a color analysis done. And so I'm a deep autumn, in case anyone wants to know.
Michael Calore: It's a burning question.
Lauren Goode: That was the burning question. I thought maybe I was a winter, but I'm a deep autumn. And so I think to date, this is the most interesting use case of ChatGPT I've experienced so far.
Michael Calore: Ever, of all …
Lauren Goode: No, that's not true.
Michael Calore: Of all your experiences.
Lauren Goode: That's not true. The other day it told me how to cook salini mushrooms instead of cremini mushrooms. I'm quite certain salini mushrooms don't exist. So it was helpful in that regard too.
Katie Drummond: Yikes.
Lauren Goode: But no, I've used it for other things too. I've used it for research and reasoning and fun things like that.
Michael Calore: That's pretty good.
Lauren Goode: So maybe later, actually, I'm going to upload photos of you both to ChatGPT and ask it to do your color analysis.
Michael Calore: I do not consent.
Lauren Goode: And then I'll tell you … OK. What color you should be wearing. Both of you are wearing all black right now.
Katie Drummond: As usual.
Michael Calore: Yeah, I was going to say. I call those days weekdays.
Lauren Goode: Anarchists in the room. I love this. OK, that's it. That's my recommendation.
Katie Drummond: I was thinking about what to recommend and it was either going to be a book, a kid's movie or food, and I was like, no, girl, you've done all of those already. You have to think of something else. I have an AI-related recommendation actually, which is, I had someone come recently to help me out with my outdoor plants, like our little backyard area in Brooklyn, and I was telling him that all of my indoor plants are dying and struggling, and I was just so confused. I was like, which window should this one go in? And what … Blah, blah, blah, blah, blah. How often do I water all of these things? I have many plants. And he recommended this app called Picture This. And you take photos of all your plants and you upload the photos into the app, it tells you what kind of plant it is, it tells you how often to water it. You can use your phone to show the app how much sun is coming in through the window, and it'll tell you if that's enough sun, too much sun, not enough sun.
Oh, and you can take a photo of the plant and it'll tell you if it's sick and what it is struggling with, which is very upsetting, but very helpful. I have many struggling plants. But it's very, very cool. It's definitely highly judgmental. I get push notifications now saying, "Don't you want to take care of your plants?" And I'm like, well, I do, but I also have a job. It's a very, very interesting product, and if you have plants, I highly recommend it.
Lauren Goode: Picture this.
Katie Drummond: Picture this. Healthy plants, healthy you. Superintelligence, picture this.
Lauren Goode: Mike, what's your recommendation?
Michael Calore: I think given the current geopolitical state of the world, it is time to re-watch Dr. Strangelove , the 1964 film by Stanley Kubrick. Dr. Strangelove or: How I Learned to Stop Worrying and Love the Bomb . It's about escalating political tensions based on a misunderstanding that leads to nuclear war. It's a farce. It's very funny. It's also very dry, but it's a great movie. I re-watch it about once a year, unfortunately. I'm usually compelled by current events to watch it. So I would say it is a good time to watch Dr. Strangelove . And I'm not saying the world is going to end, but whenever we start talking about nuclear power and we start talking about the people in the world who have their fingers on the button, it is important to remind ourselves that these are human decisions that people make about our future, and it's a great movie to help you process that information.
Katie Drummond: Wow.
Lauren Goode: So Katie's watering her plants and I'm cutting bangs and you are watching movies about the end of the world. I see who among us is really diving in and who's …
Michael Calore: So a lot of people ignore the void. A lot of people acknowledge it. I'm one of those people who puts my face right up against the glass and just scares at it.
Katie Drummond: That's just like my husband. Not to make it all about him, but … I feel like I cover the void. I do the void for my job. Every day, I'm in the void. So when I'm out of the work void, I want to go into la la land. And so last night I was telling people in Slack today, my husband was texting me links to used motorboats because he was like the safest place for us to be during a nuclear attack is in the middle of a body of water.
Lauren Goode: Good lord.
Katie Drummond: And I was like, and what exactly do you think we're going to be doing in this little motorboat in the middle of the Atlantic Ocean off the coast of New York City? I think if it comes down to motorboat or death, we are probably going to die.
Lauren Goode: Yeah, yeah.
Katie Drummond: Just saying.
Lauren Goode: That was before bed, like before you were supposed to go to sleep?
Katie Drummond: Right as I was lying in bed trying to go to sleep and the thing is like ping, ping, links to used boats.
Lauren Goode: I mean, I think we're all sort of void adjacent these days, so you can't really ignore it.
Katie Drummond: You can't ignore it, but it's a choice to lean into it in your personal time, I will say.
Michael Calore: I would say that if you're going to do that exercise, there are fewer people who it's more delightful to do it with than Peter Sellers and George C. Scott and Sterling Hayden and Stanley Kubrick. It's a great movie, so stream it tonight.
Katie Drummond: We'll link to it in the show notes.
Michael Calore: Thanks for listening to Uncanny Valley . If you liked what you heard today, make sure to follow our show and rate it on your podcast app of choice. If you'd like to get in touch with us with any questions, comments, or show suggestions, write to us at uncannyvalley@wired.com. Today's show is produced by Kyana Moghadam and Adriana Tapia. Amar Lal at Macrosound mixed this episode. Jake Lummus was our New York studio engineer. Daniel Roman fact-checked this episode. Jordan Bell is our executive producer. Katie Drummond is WIRED's global editorial director, and Chris Bannon is the head of global audio.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Fox News
7 minutes ago
- Fox News
Jennifer Lopez suffers wardrobe malfunction during concert
Jennifer Lopez accidentally left her skirt on the floor Friday during a concert in Poland. Lopez, who performed in Warsaw on Friday for her 56th birthday, suffered a wardrobe malfunction during the middle of her celebration. The "Waiting for Tonight" singer was in the midst of receiving birthday wishes during her livestreamed concert when her gold sequined skirt fell to the ground. Lopez glittered in gold wearing a metallic bra with a matching fringed skirt for the show. As she walked across the stage, concert-goers caught a glimpse of her matching underwear when her sparkling skirt hit the stage. "Thank you, Warsaw," Lopez screamed to the crowd when suddenly her skirt broke. "Oh! I'm out here in my underwear." A backup dancer rushed to her side and attempted to cover up Lopez, only for the "Let's Get Loud" singer to throw the broken garment into the crowd. "I'm glad that they reinforced that costume, and I'm glad I had underwear on," she said. "I don't usually wear underwear." Representatives for Lopez did not immediately respond to Fox News Digital's request for comment. JLo's set to perform once again this week in Abu Dhabi, United Arab Emirates. It's unclear if any additional precautions are being taken to prevent another on-stage incident. The tour, which began earlier this month, is her first tour in six years. Lopez was slated to embark on the "This is Me… Live" tour in the summer of 2024, but canceled the concerts to spend time with her family. Representatives for Live Nation explained JLo would be "taking time off to be with her children, family and close friends." "I am completely heartsick and devastated about letting you down," Lopez wrote in her newsletter. "Please know that I wouldn't do this if I didn't feel that it was absolutely necessary. I promise I will make it up to you and we will all be together again. I love you all so much. Until next time ... " Two months later, Lopez filed for divorce from her fourth husband, Ben Affleck. The Hollywood heavyweights finalized their divorce in January, with both parties waving spousal support, in addition to waiving further discovery proceedings. Affleck and Lopez were married for two years before she filed for divorce on Aug. 20 in Los Angeles County Superior Court. Her petition for the dissolution of their marriage was filed on the two-year anniversary of their Georgia wedding. CLICK HERE TO SIGN UP FOR THE ENTERTAINMENT NEWSLETTER The former couple first said "I do" during a surprise Las Vegas wedding in 2022. One month later, the newlyweds hosted a backyard ceremony for friends and family at the $8 million Georgia mansion Affleck purchased when the couple first began dating 20 years ago. Affleck and Lopez first met while working on the set of "Gigli" and were engaged in 2002 with plans to marry in September 2003, but they postponed the wedding and ultimately called off their relationship at the beginning of 2004. "Bennifer" reunited once again before the summer of 2021, and Affleck proposed to Lopez in April while she was taking a bath.
Fox News
37 minutes ago
- Fox News
Prince William brings surprise guest to international soccer match in sweet father-daughter outing
Prince William brought a surprise guest with him when he went overseas to watch England's women's national team compete. The Prince of Wales was photographed alongside his daughter, Princess Charlotte, as the two of them sat side by side during the final of the UEFA Women's EURO 2025 in Switzerland. They supported England's women's soccer team, nicknamed the Lionesses. This occasion marks Charlotte's first overseas soccer match. A photo of the two of them at the match was posted on the royal family's official Instagram page, and fans were quick to point out how similar the young princess looks to her father. "Charlotte takes after her dad so much! The genes are strong ❤️ looks like a lovely day, go lionesses!" one fan wrote. "Totally copy paste from her daddy 😍," another commented. Others pointed out how great it was to see the father-daughter duo bonding, writing "Such a beautiful picture and it's wonderful to see Charlotte with her dad 😍👨👧," and "So sweet! Father daughter moment ❤️👑❤️." While this was Charlotte's first overseas soccer match, it was in no way her first time attending a sporting event. In August 2022, the Princess attended the Commonwealth Games alongside her parents, marking the first time she made a solo appearance with her parents, without her brothers there. At the age of seven, the princess stole the show when she was spotted making several funny faces among laughing spectators. More recently, Charlotte attended the Wimbledon men's singles finals with her father, her mother, Kate Middleton, and her brother, Prince George. She has attended the event in the past, sitting in the stands alongside her mother in July 2024. Prince William shared some insight into his relationship with Charlotte in April 2024 when he visited St Michael's Church of England High School. When speaking with students, the proud father shared his daughter's favorite joke. "I am here with the gang here talking about male mental health, and I've been asked to produce a dad joke. I was trying to channel Jack Whitehall because most of his jokes are pretty dad-like," he told the students, according to reports. "I think what I'll do is say a joke that is Charlotte's favorite joke at the moment, that she keeps telling me." He went on to tell a knock-knock joke. When the students asked "who's there," William responded with "interrupting cow," and before the students could finish asking "interrupting cow who?" William interrupted, saying, "Moo!"
Vogue
an hour ago
- Vogue
Prince William and Princess Charlotte Share a Sweet Moment Cheering on England's Lionesses to Euros Victory
When England failed to defend a goal against Spain, both William and Charlotte showed visible distress, with William leaning in to share some words with his daughter. They talked animatedly with their Spanish counterparts during the game, enjoying the light rivalry which kept spirits jovial. When the Lionesses sailed to victory, they hugged each other and jumped for joy. Princess William wore a dark suit and subtly patriotic tie, navy with thin red and white stripes. Taking a leaf out of her mother Catherine, Princess of Wales's style book (who, when together, will often share lovely mother-daughter dressing moments), Charlotte opted to re-wear a frill-shouldered, navy polka-dot dress from Guess, which the young royal wore to Wimbledon last year. Kate Middleton often delves into her own archive of elegant tea dresses and structured tailored looks. When she joined the rest of the British royals for the Order of the Garter service at Windsor Castle most recently, Kate pulled a Self-Portrait dress from her wardrobe that's proven to be quite a favorite. She'd worn the dress for both the 80th anniversary of VE Day and the Platinum Jubilee in 2022. It looks like she's passing that mindful sartorial sensibility down to Charlotte.
