Episode 592: AI and Relationships: The Future of Connection, Dating, and Loneliness

0:01Hi guys, it's Tony Robbins. You're listening to Habits and Hustle. Crush it. Welcome to another solo episode with me, Jennifer Cohen on Habits and Hustle. And we have a treat. We have a treat because you guys have not met my friend Hamet Watt, but you're going to meet him now. So just some background, you know, Shani used to do a lot of these foils with me. She was my foiler and I would do some solos and she would kind of like banter with me. What's a foiler again? You. Okay. Basically someone who I can like shoot the shit with basically and like kind of like go back and forth with. And I really liked Shani because she was A, she's super smart and she has like interesting perspectives and she can challenge me.

0:48And she's just, I like her. She's like a friend of mine. Got it. But she's now gone to Israel for I think an unknown, like an unknown, like an amount of time that is just not... Forever, basically. Basically forever. We don't know yet. It's indefinite. How about that? Indefinite. Got it. So anyway, I'm actually introducing you as you're talking to me. So now I found another potential foiler who I get along with swimmingly. And his name is Hamei Watt. Did you say swimmingly? Yeah, swimmingly. I never heard that expression. You haven't? No. So we get along swimmingly? Like we get along well. Like we interact well.

1:27We've got a good dynamic kind of thing. I've known Hamei for 10 years. About that, yeah. 10 years, okay. And I think we've always clicked from day one. We always had a nice little banter. Hamei is a great guy. You're going to see very quickly why I like him so much. He's got a very good sense of humor. He's very smart and sharp. And he brings a really good perspective. He's a founder and he's an investor. He's had a lot of success in his professional and personal life. He used to run a fund called Upfront Venture. Well, I was a partner at Upfront. A partner. Okay, excuse me. You're a partner. I didn't run it.

2:03I didn't run it. What's the difference? Tell us. There's plenty of people that run the firm. I was just one of the folks that was investing and working on companies. That's it. Okay. Nothing too special. Well, what do you mean? If you're a partner, weren't you one of the people who ran the fund? Well, actually, in most venture funds, there's managing partners that typically run the fund, and there are partners that might do investments. Oh, so you were below the managing partner. Okay. But now, that was like back when, now Hamei has his own fund. Can I say who's backing it? Yeah. I could. Okay.

2:33So Hamei is like a superstar. Really, he is. He's got a great eye for innovation. He runs a human performance lab and is a part of creating some of the coolest brands and products right now on the market. Thank you. No, it's true. And he's very, he's like an expert in AI and human performance. And he's so good that Amazon and Google are backing his fund, which is pretty impressive. Yeah, it's exciting. It's exciting and super impressive. Yeah. Enough about me. Let's get into this stuff. What do you think of me? Like, enough about me. What do you think of me? So anyway, that's who Hame is. So just now when you see him, that's who he is.

3:15So welcome, Hame, to the podcast. Thank you very much. Thank you for being here. I'm honored to be here. Oh, the honor is all mine. So thank you. So since we've never done this before, I figured I would pick a topic and then you and I can just like riff on it for like 10, 15 minutes. Okay. Okay. And the topic, the first one I want to really kind of go over with you is, I mean, since you are a special, like an expert in AI, I think we should start with something like a softball. Okay, good. Okay. Good week to be talking about AI. Why? Tell me. Well, last week there was a bunch of announcements.

3:50I'm sure you saw the OpenAI announcement that they - Tell us, because maybe some people have - This is exactly what I want you to do. Oh, good. Okay. So tell me, is there anything - What's new on the horizon for AI? Well, OpenAI announced a new model. They call it Astra. Their sixth model, I think. Yes. And it's scored like 100 on the benchmark for AGI. So they're basically going to have to rethink the whole scale for how you measure how great these or how strong these AI models are. What do you mean it scored 100? What does that mean? Well, the average human, I think, scores around a 40 on this scale in terms of intelligence, this benchmark for intelligence, and this scored 100.

4:29So people are saying this is very close, scary close to AGI. And if you've played with it, it's insane. It's just insanely good. How is it, if this is ChatGPT, which is open AI, does that make Claude and all the other ones now less effective? Yes, but Anthropic— Because they're always competing, right? Yeah, they're always competing. And Anthropic is going to announce something in the next week or so. So they're going to one-up each other each time. Can they go higher than 100? That's the question. I mean, there's potentially a new scale that they have to develop to figure out what that intelligence level should be.

5:08But right now, this most recent open AI model scored at 100. So, okay, so which one do you, like of all the AI platforms, which is your favorite? Well, we have our own harness and platform at Share. So we use all the models and it really depends on what we're building and what we're using. Okay. But personally, I had probably been defaulting more recently to Claude, although I'm now kind of spending more time with OpenAI as of a week and a half ago. So it's - Because of this new thing? Because of this new thing. And it's crazy because also, I think it was either the beginning of this week or maybe the end of last week, two Anthropic employees left Anthropic because they said, we're getting dangerously close to AI that could wipe us out.

5:54Right? We're getting dangerously close. I think they said there is almost a 10 % chance that AI could wipe us out, which I don't know what they meant by that. They're reasonably senior folks at Anthropic. They left under some sort of swarm of controversy. but it's kind of a scary and exciting time at the same time. I tend to be an optimist, so I don't think it's going to wipe us out. How can it wipe us? It's basically like machines are going to take over humans. This is literally what the sci-fi movies were talking about. Yes. And now the people who are the humans at the sci-fi companies or the AI companies are actually scared of what they, they're smarter than the people who built it.

6:31Yes. They are smarter, these machines. We are approaching a place where the AIs are significantly smarter than the humans that have created them. How does that happen? How much time do we have? No, no, this is actually, I think a lot of people are very, like this is, before we can do another hot take with something else, but I think it's really interesting. Like, how does that happen? How do you build something, a human being build something that ends up being a million times smarter than what the person built? You know what the analog is? You have kids. So imagine if one of your kids is a genius and you've trained them and you've exposed them to something, they could quite conceivably be smarter than you at some point.

7:13At some point now. I can't even do fifth grade math. Exactly. No, my kid is already smarter than me. So, yeah, so I think that that is kind of how it's happening. And these models are getting so good at training on all of the world's information. Yeah. That, you know, there's no human that can take all the world's information in our brains and remember it and know how to reference it for every task imaginable. That's what these models can do. And when you start adding more compute and more data and more types of data, video, obviously all the text, anything that's been written on the Internet, all the imagery, all the songs, all that huge corpus of data means that it's trained on basically everything.

7:56And as the models get stronger, as the compute gets more robust, you have an ability to do some pretty amazing things. Right, but also scary things. So why, if these employees left Anthropic, why? They didn't talk about why. But are you sure it's because the machine just got, like, what was the reason that they actually said? What they said was they feel like things are moving too fast in the world of AI. Yeah. And that we are approaching a place where the risk to humanity is meaningful enough. What did that mean, though? What does that mean? What could happen? Yeah. What could happen? Everything from the Terminator.

8:37Yeah. No, that's exactly the movie I was referring to, by the way. AI start to optimize for themselves over humans. Again, I don't believe this. How do you not believe it? Don't you remember, like, not that long ago, there was that story about some guy, the AI bot was threatening to tell his secrets. Do you remember? Which one? No. Exactly. Which one? I'm talking about Anthropic Claude blackmail experiment. It's that great hot take talk. Okay, Anthropic put Claude in a simulated company environment where it had access to fictional employee emails. The AI discovered two things. An executive was having an extramarital affair, and that same executive planned to shut the AI down at 5 p.m.

9:27And in some test runs, Claude decided to use the affair as leverage and sent a message essentially saying, cancel my shutdown or I'll expose you. You didn't hear about this? That's scary. You didn't hear about this whole thing? No, I didn't. Yes. The important distinction, this did not happen to a real employee in the real world, but the executive affair, the company, and scenario were fabricated as a safety test. Scary. Yeah. Researchers deliberately created a situation where the model act that where that happened, that can actually happen. So that's really scary. It is scary. I think the good news is like these guys, there are a lot of people that are putting safeguards in place to make sure these things are less likely to happen.

10:15Doesn't mean they won't happen. But I think when I do the math, I think of the benefits that AI can have as it relates to health, all the drug discovery that's happening, all the things that are going to be amazing for more efficiency, effectively bringing the marginal costs of goods way down so things are much cheaper to afford. Like you got to trade that off relative to the jobs and relative to all the downsides. So when I do that math, if we're playing enough offense to do enough amazing things with AI, we're going to be okay. because the things that will... You're saying the benefits outweigh the negatives.

10:49If we're intentional about it and we're really pushing hard around those opportunities. Okay, so out of the platforms, can you kind of give us some indication? What is ChatGP? What do you use ChatGPT for? What do you use Cloud for? Because what I've noticed is I used to use ChatGPT. Then I went to Cloud and was only using Cloud. And then I realized recently in the last month or so that Claude became better. Not Claude. ChatGPT became better. So what Claude used to be good at, I don't find they're as good at anymore. Yes. Like the writing and the personal questioning. Yeah. A lot of people have talked about Claude becoming less good than it was.

11:35Really? Yeah, even though Fable 5.1, which before more recently last week, was just the best model known, especially as it relates to building. Was it 100? No, it was not, right? I think it was in the 70s. That's what made this such an important and crazy announcement. Now, who knows? The next announcement may mean Claude is on top again. Both of these are trillion-dollar companies, near-trillion-dollar companies, about$800 billion apiece, that are trying to go public. So they're right now showing us their best they can because they want to make sure they go public in a big way and they're successful.

12:10Okay, so what do you use what for? Just give us that and then we can move on. Well, I use primarily, I use our own at Share. We have our own harness, our own platform that we use. And that chooses, yeah, it chooses whatever model is appropriate for the job. And we have a router that helps us do that. So you guys create your own AI? Well, we create our own platform that leverages a lot of other models, frontier models, including our own to help us with what we call innovation intelligence. Our lab, we're starting new companies. We're running experiments constantly. So we're using that platform to do that.

12:43But personally, I probably use GPT more because I love voice. Don't you like talking to yours? Yeah, me too. I like it. That's what I like, too. I think my voice is like an African-American woman, too. She's kind of— Really? Her sound is—it kind of just—it soothes me. You can change it, too. Yeah, you can change it.

13:05I want to take a quick break from this episode to thank our sponsor, Therasage. Their tri-light panel has become my favorite biohacking thing for healing my body. It's a portable red light panel that I simply cannot live without. I literally bring it with me everywhere I go. And I personally use their red light therapy to help reduce inflammations in places in my body where, honestly, I have pain. You can use it on a sore back, stomach cramps, shoulder, ankle. Red light therapy is my go-to. Plus, it also has amazing anti-aging benefits, including reducing signs of fine lines and wrinkles on your face, which I also use it for.

13:47I personally use Therasage Trilite everywhere and all the time. It's small, it's affordable, it's portable, and it's really effective. Head over to therasage.com right now and use code BEBOLD for 15 % off. This code will work site-wide. Again, head over to therasage, T-H-E-R-A-S-A-G-E.com and use code BEBOLD for 15 % off any of their products.

14:27So what about, besides ChatGPT and Claude, where are the other ones that are Gemini, obviously? Does anyone really use Gemini? Gemini's been trying to improve, but they're just not competing as much in terms of the consumer. How? Why not? It's like Google. They're Google. They invented this, too. I know. How can they not compete? You know those two guys came back. I know. They have to. They're trying to compete. I know. They had to come back because they're like, we're going to die. You know Larry and Sergey, right? Yeah. Larry and Sergey. The guys. The guys. The best friends. Yes. Yeah, but like, and it's not working yet.

14:58Not yet. It's competitive. How far away? Who else is there? There is some discussion that the newest, the model that will be coming out, Gemini's new model, will be very competitive and on par with the new chat GPT and Claude. So the three of them are really duking it out. And then you have, you know, Grok, which is, you know, X's AI platform. I don't even understand the Grok one. He sold it to, I just listened to an interview with him. Well, he merged it all with SpaceX, right? Well, he went to Naveen. Do you know what it was that? Grok was sold for like$3 or$4 billion, I thought only. Did you see this thing?

15:38Maybe you're thinking about another Grok because there are a couple Groks. There's more than one Grok? Yeah. Everyone thought that was the best idea for a name of a company at the same time. And I think there are two. Elon's AI company, Grok, is obviously an AI. And then there's another one that's in the inference game. So yeah, it's a little confusing. You're probably talking about the other one that's inference. That's just like compute capacity. capacity what does that mean well it's i think it's an alternative to i think the one you're thinking about nvidia bought is the inference yeah nvidia now part of the uh nvidia capability and hardware capability so if you want a data center and you want compute and you want inference then that's where you're spanish i don't know what you're talking about does your audience care about this stuff or do they want to care well no i i care and i think someone cares because if you don't care you should it's like what you should it's like this is what what the world is morphed evolved to.

16:28It's crazy. Like if you don't know about this stuff, guess what happens? You're going to be, you're falling behind. Yes. Like, you know, it's, it's actually like a thing, you know, like we don't have a choice anymore. If you don't, even if you're not interested, get interested. Yes. Right. Like totally like people who don't know how to use this stuff. It's not like it's Facebook. You can get away with not knowing how to use Facebook, But literally, AI is taking over every area of your life. It is. And in some ways, it's the most empowering, most amazing thing for humanity. We can cure diseases.

17:02We can solve problems that haven't been solved in years. In fact, there was another announcement today that this new open AI model cracked some scientific code around fluid dynamics that they'd been working on, that people had been working on for years. and this model kind of cracked this mathematical equation that they think is going to impact all kinds of things. And I actually just saw this out of the corner of my eye. I didn't read it in detail, but it's pretty amazing. This happened today or yesterday. But can I tell you something about humanity that I think is interesting? You asked how AI might take us out and everyone talks about the Terminator and all this crazy stuff.

17:39You know what I think it might be? And no one's talking about this right now. The sex rate is going down. The birth rate is going down. The number of people in relationships is actually going down. Relationship satisfaction is going down. All these things are not really great. So what happens when the birth rate falls below a certain point? In some societies, it's already fallen past that point. I think South Korea, it's actually perhaps at the point of no return now. They're having to figure out how to get more people in to increase their population. So what could happen? okay, the birth rate, I don't know what it is today, maybe 2 or 1.9 or whatever, goes down to 1.4, goes down to 0.75.

18:22All of a sudden, you have a beautiful robot that you can procreate with, and that just compounds it even more. You can't procreate with a robot. I spoke too soon. I was going to say, you can procreate with it. You cannot procreate with a robot, obviously, but you can fornicate with a robot, which might take that sex rate even down further. Okay, well, that to me is where the world is going. Have you heard about this whole thing about for the first time ever, I heard Esther Perel on a podcast, on Oprah actually, saying that she recently had a patient who came with his chatbot to a session. To have a counseling session.

19:02Yeah, that's what's happening. So to your point of there, that to me is where it's becoming a very, it's, I think this is where the point of no return is happening. Because people are now would prefer to quote unquote date a chatbot because you can program it to say what you like, act how you like, sound how you like. Respond to all your needs. You respond to all your needs where then a real person can't compete. This is also the problem with, if you haven't noticed, there's influencers now online that are like AI, and people don't know the difference. So now you're, who look fantastic, they're 80 years old, and they look like they're 30, and you don't know the difference.

19:53And people are following these people, until you're competing with something that's not even real. What happens when that goes into the real world? It is in the real world. Right, but when you have, right now, you have a digital equivalent of that, and the digital equivalent that you can actually have a relationship with, what happens when you can also have a physical equivalent of that? How do you do that? Is it virtual reality? Like, how would you do that? Oh, eventually robots, for sure. Yeah, but it's still a robot. Yeah, but eventually the distinction between a robot and a human is going to be so thin.

20:29The same way today, you could have a relationship with an agent, with an AI agent, and the distinction there between a call with a real human and an agent, not that much. The gap is getting more and more narrow. Eventually, if you play this out without being conscious of it, it's kind of scary, right? AI doesn't have to kill us. We just kind of naturally sort of fade out. Well, that's what the scary thing is. That's scary. Do you know anybody who is actually using these agents or bots instead of like real people for relationships? There was an incredible story. I think we can maybe find this and put this in the notes.

21:06I want to say the company was called Replica. Okay. And they had a virtual boyfriend, virtual boyfriend, girlfriend type of thing. And it was adult as well. It would handle adult conversations, right? Sexual conversations. They decided to turn that adult feature off, right? So you couldn't have a sexual conversation with your virtual boyfriend or girlfriend. They had a class action lawsuit. People were up in arms that their virtual boyfriend or girlfriend was basically lobotomized and wouldn't be sexual with it. No way. So you should look this up. I'm not making this up. So, yes, there was, I want to say there was close to 100 ,000 people.

21:49It was a significant, and there was thousands of people. that were up in arms because their relationship, their primary relationship, which was not real, had fundamentally changed because of the way the company changed their policy. No way. Yes. Like, what kind of person, though, does this stuff, though? Have you ever talked to one of these before? By the way— You should. We should call one up right now. We should. How do I find one? I mean, you could probably go—actually, you can go on any of, like, if you go on Grok, not that I've tried this, you can you can some of my god you have tried this of course i've tried it i've been trying siri since she first came out and siri's very appropriate i've been trying her pushing her to the limits trying to see if she would yeah talk dirty to me she never does really very responsible very respectful but some of these let's just say they're very very very very bad really yes okay how do I find it?

22:50What do we do? How do we find one? We're not going to promote these. No, I want to know because my next question, she was like, who's actually doing this? Millions of people. Well, I know what I'm saying, but who are, what's the, what's the. You know what the surgeon general said the number one health concern was? What? Loneliness. It is. Well, that's 100%. Yes. But, but see how this works? Because you're lonely, you, gravitate to these agents, right? But because now you have an agent, then you don't gravitate to human beings. It's like this really bad loop, right? It's a very, very, it's a scary loop.

23:28Can be, I would say. And in most cases it is, you're right. I think the optimist in me says, well, maybe there's so many lonely people, maybe they start to exercise the muscle around connection. And the optimist in me also says, well, maybe some of these AIs will start to teach us as humans how to better interact with other people. No, we're going to talk about all that. That's like a whole area that I want to talk to you about. Yeah. The human connection piece. I think we'll talk about that in another podcast because we're going to. But just to wrap this up for a second, how do you find an agent or a bot?

24:03Like, how is it different than going on ChatGPT and talking to your, like, hey, tell me, you know, the weather in Belize. Like, is it not that chatbot that's talking? It is very similar, although some of the chatbots, including, I think OpenAI has made the announcement they're going to go adult, which means that, you know, the not safe for work type of conversations can happen even on, you know, big applications like ChatGPT. but there are tons of virtual boyfriend and girlfriend apps out there. And that's, those are the guys that are really. These are not on the base. Those. So the chat bots and agents, you can't find on a cloud or a chat GPT or a grok or.

24:41No, but many people are building using those tools, but they're building them as separate, you know, separate companies, separate services. Before I, before we wrap this, I want you to give me like, tell me what happens when you did it. So tell me what happened. Well, I've just to be, yeah, be honest. I want to know. I'm married, happily married. Yes, we know you're happily married. But you're curious. However. I'm just teasing. No. There are times where, like, when you have, it kind of goes to a basic human instinct. Ever since the first agent, Siri, the first AI, and now Siri's responding to me, but the very first time, I tried to see how far I could push her.

25:25Do you know what I mean? I would try to say, oh, what are you wearing? And she would say, that's not appropriate. Siri would say that? Oh, yeah. She would very tastefully redirect and say, that's not what we're here for, right? Okay, but I want to know what happens when you talk to these agents. They will go as dark and as dirty as you'd like them to be. Tell me the dark and dirty that happened when you did this and where you found them. Not me, but a friend. Of course. Always a friend. Always a friend. A friend was telling me that you certainly can have phone sex with them. You can? Yeah. Like, give me an example.

26:05What kind of show is this? I want to know, like, how it starts. Do you have to pay? Like, is it free? How do you start it? Like, what's the process? The AI agent boyfriend and girlfriends out there, for the most part, they have a beautiful funnel where you would start talking. I honestly have not done any paid things. Do you have to be a member? I was going to say, how does it work? Do you have to sign up? The ones that are out there have all sorts of various sign-up approaches where you can talk for a little bit and then eventually they hit you with a paywall and you go. But even Grok, which is largely free, Grok has a not safe for work version.