Mark Cuban warns that ads inside AI responses will be manipulative and erode consumer trust

likely doing in the world of advertisements. We're going to debate that. Welcome to the stream. Mark, how are you doing? Good, guys. How y'all? We're doing great. Thanks so much for taking the time. Good to meet you. Uh where should we kick it off, Jordy?

Do you dive straight into the debate over artificial intelligence and advertising? Um sure. Maybe you could set the stage for us. What What set off your alarm bells? What triggered your initial uh worry about this idea of ads in LLMs? Because this isn't the internet, right? It's not just a normal digital platform.

It's it it's a platform that depending on how it's trained, depending on what u pre-prompts you put in, it could be very manipulative. And there's not really age limitations at all. There's not really any limitations whatsoever. And normally I'd be, let's just be lz fair, right? The market is the market.

But you've already heard, you know, of stories of kids going down these rabbit holes and, you know, other people being told, you know, to kill themselves and you just you just don't know what can happen. So if I if if I build a model and it becomes popular, even if it's only a niche model, right?

Just like there's headsp space and there's calm. Let's just say I build the calm AI large language model and I have millions and millions of users and I want and I have an advertiser who sells Xanax or sells, you know, whatever, right? And they're trying to get more sales and I need the money. Mhm.

I can easily easily train this thing to manipulate people into doing things we typically would not allow people to do. And that's my concern. And so if we just want to put up, you know, um just little ad, traditional graphics, JPEG ads, all you know, put them into um the chat list on the left hand side. Great.

You can do all the display ads you want. You could do video there all you want. But if it's in the response that comes back from the model, boy, you're you're going to have lots of problems. Yeah.

I think I think Sam Alman said something to this effect, saying that uh artificial intelligence will be superhuman at persuasion before it is truly super intelligent. It will be very good at convincing people. So it is reasonable.

What do you think of uh Sam Alman's uh latest commentary on how OpenAI might get into this world that the actual LLM interaction, the chat would not be uh would not be monetizable, but uh if you if it made a recommendation to buy a product and then you said, "Hey, I want you to go check out for me.

" It sends out, right? Referral fee. Is that okay? Yeah, it's already doing it. You know, it's this, you know, it's one thing to have the model itself convey and try to manipulate. It's a whole another thing when there's a search query. Yeah. Right.

So, we get a lot of traffic from um um OpenAI Chat GPG for cost plusddrugs. com. In fact, I've used it because sometimes if I want a price for one of our drugs and I I don't want to take the time to look at the latest price list, I'll just put in what's the price for, you know, Darren, whatever it may be.

And um it's right there. And so yeah, of course they can extend that, but response is a whole and models are are smart enough to know what the difference between a response to a query versus influencing and engaging in a chat. I guess the question Yeah.

I mean, I think I I I think to to double down on what you're saying, an example, uh, if you're going from search, which is people coming in with a specific intent and and maybe doing some product research and you have ads and you have, you know, organic listings or or web pages, that's one thing.

But if if if uh various AI models are seen as trusted advisors, uh, if you compare that to a situation, if I ask John, what kind of car should I get? And John starts giving me answers and saying uh like I think you should I really think you should get this car. It's really nice.

And then I buy it and then later I find out John was getting a 10% kickback. I'm like I hopefully I like the car but if I didn't like what are you doing? You know and this happens in a business context too. You might you know people get in hot water all the time because they're taking referral.

Yeah, in this particular case it's really easy just like you know there's you know emoji for um sources or just says sources you could put a dollar sign just to show they're being compensated. Sure. Sure. Sure.

And do you think that needs to happen at like the FTC FCC level or are you do you think I don't know the legal aspect of it enough? You know hopefully it's self enforced but you know there's going to be idiots and who don't self enforce and but particularly for medical stuff.

I mean, it's easier to talk to chat GPT about your own personal medical situation than it is to find a doctor to talk to sometimes. But but does that call example you gave matter? Because uh like I I I imagine that there'll be a power law outcome here.

And yes, I could set up a search engine that serves that serves ads and doesn't disclose that they're ads. But how am I going to take market share from Google? People are going to go to Google and Google does say when it's an ad.

And so like as long as the power law winner is a good actor, we kind of get there's also an interesting there's also an interesting thing where there's this self selfinforcing sort of positive effect which is if if people rely on chat GPT for example to help them do product research and buy the right products and you buy a bunch of bad products in a row because they're just pushing the the product that that they're getting paid the most.

Eventually I do think that consumers are smart enough to realize I'm not going to use this app anymore. I I wouldn't agree with that, Jordy, because I think people are primed for misinformation these days. You see it in healthcare continuously.

You know, you saw it with co vaccines, people on their deathbed saying, "I wish I had taken the vaccine, you know, and you see it now with, you know, um raw milk where now there's more people who have, you know, there's examples of people who have died and people just don't care. " Yeah.

you know, they just go on and maybe they'll say, you know, chat GPT is woke, so I'm using right GPT. Yeah. And that's because those are the kinds of answers I want because, you know, there are people who want to be influenced and want to be part of a group when it comes to the product decisions.

Just like we saw the anti-Bud light and that kind of builds and you see, you know, pro right-wing products. I just don't know that it is an efficient market enough where people are going to take responsibility for just making a bad product choice and then going to another source.

There's just so much personal identity and how we use technology. That's part of my concern. How how worried would would you say you are about AI broadly?

I think over the last few months the industry has woken up to uh you know a year ago people were kind of laughing about AI safety right and and thinking about these fast takeoff scenarios.

Now I think people are kind of waking up and worried about the general well-being if you just give hundreds of millions of people you know unlimited access to these tools. You know they are gonna that there there is going to be uh some effects.

Some people will be good some people will be bad but but what's your what's your high level read? Well it's changed a lot. We went from fear of the Terminator Yeah. to fear of being influenced in ways we didn't expect because they're such good communicators. Right.

We can't talk to other people or we feel you know shy or not open in communicating with other people experts or friends or otherwise but this is our best friend who never gives us a hard time and depending on how it's programmed will compliment what a great question Jordy what a great that is that is a really interesting approach so there's a lot of things to be concerned with there particularly because AI is not smart yep right AI is just statistical right now.

And I'm of the mindset that that's not going to change for a long time. You know, one of the reasons that you don't see full self-driving cars that are just everybody using no matter what is because it's it's easy. There's so many adversarial things that can happen to a car. Yeah.

You know, we have um Australian many Australian German Shepher or a million Australian Shepherd, right? Tux and about six years old. I trust Tux at an intersection to know when to cross before, you know, that he's never been to before before I trust a car.

And that's the whole point with AI because there just rules and laws and latencies that they that it can't account for. It's all text, some graphics, some video, but it's always behind. And then that's part one. Part two, we're I think we're going to see a ground swell in how um intellectual property is dealt with.

A ground swell of change and how IP is dealt with because there was for a long time it's been publisher perish. Get your stuff out there. Get it into journals. Make it available so that you can present yourself as being an expert.

Well, the minute you do that right now, it's going to get absorbed into, you know, training data for these models and you've lost everything you've just created.

Well, what if I want to what if I'm taking the side of big raw milk and I want to and I want to get people and so I should just write as much as I can about that's what you're going to do, too. You're people are going to use bots to try to influ to post things to try to influence like the um robots.

ext for AI, you know, for AI. You can you can structure it in ways so you can set up sites where you can tell it exactly where to go, you know, tell it all why raw milk is amazing, right? But the bigger point is AI doesn't understand death. AI doesn't understand pain. AI doesn't understand context.

AI doesn't know what happened two minutes ago. It doesn't know that there was a plane crash as it's about to tell you or something happened, right? AI can't tell you why that ball fell off the table unless you tell it, you know, this ball fell off the table, what happened next, right? It's just not smart.

And as long as it's not smart, it a can be influenced and b, you're just not going to know what it's going to do next because it can't it doesn't have judgment. So, what would you tell the folks at OpenAI and the other foundation model labs on how to structure their business?

It feels like a firewall between there needs to be some sort of truth seeeking organization that's focused on on delivering the highest quality results and then on the flip side you have a monetization team. No, I mean you can't check for quality. That's impossible. Yeah, of course you can.

You like you can you can always like benchmark these things or just like vibe test them with the with the with the audience. You know, Jordy was giving me I was giving Jordy online. podcast are going to be completely different. Right?

But what I what I'm saying is like is like Google Google does try and try and deliver the highest quality search results for a particular query and then separately they have an ad auction uh function that that adds as an ad platform and those two teams are separate and so like if your age if your age is showing up as one of those like knowledge boxes on Google you can't just pay to have them change it like that's a separate business line and it's a separate team and they are separate products in the UI.

I yes they exist on the same page but they're separate and that's fine right I mean to me the bigger questions are what's the value proposition for AI as we understand it today or at least as I understand it today right I don't see AI getting smart I look at AI and I think it's the world's best library let me explain it another way yeah we all have that one friend that remembers everything Yep everything right What did we do that one day back in college?

Remember when you know our one buddy, oh yeah, this is D. That's AI. They'll it'll remember everything. It'll give you the, you know, if you give it A, B, and C, it'll figure out D, E, and F, right? But it doesn't even know XYZ exists, just like you're right. So, it is just like the world's biggest library.

It is just like a meeting of of the minds of of just regurgitation.

And so knowing that then you you have to there's going to be so much competition between the models that is the biggest driver right now in terms of revenue generation right because there's not there'll be five six whatever it is foundational models before there's consolidation but beyond that there's going to be millions of models millions and millions of models and every brand is going to need a model particularly like in healthcare like if you are Stanford medical right if you are Mayo Clinic if you're MD Anderson you're not going to just you know put out all your IP from the doctors and researchers and scientists you're paying so that chat GPT can absorb it and have the benefit of all that knowledge so it remembers everything that that all your doctors and scientists did.

You're going to keep that to yourself. And so in terms of the business models, these guys are going to have to start making decisions. Are they start going to start paying a lot of money for content? Because as much as they're investing in the technology, they've reached a lot of limits, right?

You don't you're not hearing about, wow, these guys are getting to do these new topics that I had no idea that they could cover. They're paying for a ton of content to be clear. I mean, they pay Reddit, they pay the New York Times, they pay all sorts of Yeah.

We had we had a group reach out wanting to buy our back catalog. I was like I assumed you already scraped it. But but think of it if you're I mean let's talk about it from the government perspective. Sure.

So they just cut all these NIH grants right all this stuff that presumably went and created new drugs which they have foundational research basic science all this research. Yep.

If the government really understood AI, they'd be like, "Okay, we're going to keep on investing in this research and then I'm going to make it available to the biggest bid, the highest bidder, whoever it privatize it. I like that. Make some money off of it. Not even privatizing it, but monetizing it. Monetizing it.

I like this. This is good. Make some money off of that nonprofit research because that's that's the way things are working right now, right?

And I don't and I don't have honestly I don't have a problem with that at all if it gets us more solutions but or doing our own you know the government doing its own model and making that available for a fee. Yeah.

Those are the decisions from a business perspective that all these models are going to have to make otherwise how do they keep on adding more training data because there's no way to synthesize all that You're not synthesizing.

You're not creating syn synthetic data that all of a sudden is going to, you know, you could tell, you give it all the backstory you want that you're a doctor, that you invented this, that you did that. It ain't going to invent what the doctor and the scientists are going to invent. Yeah.

At any point during the AI hype cycle so far, did you think that this time might be different? There was I would say a year ago the common a common line was this isn't like the internet bubble because there's real revenue.

You know these companies are ramping revenue to hundreds or billions of dollars of revenue very quickly. This is different. This time is different.

Meanwhile, uh and and I wouldn't lean on this too heavily, but MIT had a report this week saying that, you know, 95% of of Gen AI pilots in the enterprise are not turning into uh not creating real value. I'm curious what your read has been.

I've been through every single technology, you know, event and evolution and this blows them all away. Mhm. Now, how you implement it in business is a whole different issue.

Like literally, when I was 24, I was walking into companies who had never seen a PC before in their lives and explaining to them the value and having these guys going, "Well, son, I got this receptionist right there. I got that secretary. I'm never going to need that ever. " Right.

And but then you know my business then was helping them figure out how to implement it to give them an advantage. They're going to be integrators that particularly young kids like when I'm telling my kids who are 15, 18, 21 or 19, 21 and kids going into school, what should I do? What should I do?

I'd be like AI research. Learn all you can about AI but learn more on how to implement them in companies. Right. Because to your point, companies don't understand how to implement all that right now to get a competitive advantage.

You got the head of Microsoft saying software is dead because everything's going to be customized to your unique utilization, right? Or use usage. Who's going to do it for them? Particularly small to medium-siz businesses. There are 33 million companies in this country. 30 million of them are soloreneurs, right?

Um single person enterprises. There are only, you know, there are millions of companies that have one, five, 10, 50, 100, 500 people that aren't going to have AI budgets, aren't going to have AI experts. This is where kids getting hired coming out of college are really going to have a unique opportunity.

If you're spending your senior year in college right now, your senior year in high school even, whatever it is, your excess time, and you're learning the difference between Sora and VO, and you're learning how to do all this video, you're learning how to customize a model so that then you can then walk into a company and say, "I understand your business as a shoe company selling shoes at a retail store.

We're selling and selling shoes online. Let me show you how to benefit you. " That is every single job that's going to be available for kids coming out of school because every single company needs that. There is nothing intuitive for a company to integrate AI and that's what people don't understand.

We hired two we hired two interns this summer at the end of the last school year because they just built products. Instead of saying, "Hey, here's what I can do. " They just showed us they took a day and just built something. We said it day, right? They went to chat GPT or perplex. Yeah. Yeah.

Well, the bear the bearcase John was joking about this though. I thought this was interesting. Uh we we we had one of our interns built this product. It's tbp. com/guest. It's this guest directory transcript tool. And he built it in a day, but then he had to spend the whole summer improving it.

So that was that was a bullcase for software engineers still having a job. But you don't even have to be a software engineer, right? You just have people are afraid to ask the models the right questions. They're afraid to ask complex questions because they just presume that it can't they can't answer.

Kids coming out of school today that are fearless in the questions they ask and the follow-ups and their ability to prompt. They have more skill than everybody in every major corporation with under a thousand employees. Right. So, I completely agree with this. Jobs for everybody.

This feels like a It feels like you're making the argument for putting ads in AI and keeping it free so that kids can have access to frontier models and then go and and explore, learn, implement things at businesses. If all of a sudden you can't get in the game, you can't even interact with a cutting edge.

Well, Mark saying he's cool with display ads. He's cool with a Laboo showing up in the sidebar. Yeah. On the side, chat. That's fine. But think of it, you're Okay, there's a death war going on with the Frontier models. Yes. War. Yeah. It's zero sum game in their minds. Not all will make it. Some will get consolidated.

Maybe one goes out of business. Yeah. Right. And so for the next however many years, however long, there used to be a time when Excel costs $499 to buy a spreadsheet. Now, you know, not free, but like basically Google Yeah. Google Sheets, right? Google Docs, etc.

The basic models are always going to be free because of ads. Like the reason Google Sheets is free is because of Google's ads business. When was the last time you Google? No, you don't.

But but the reason that they can subsidize it is because they have this amazing ads business over here that throws off so much cash they don't care about monetizing it. Yeah. But in the case I mean to to to uh to to take Mark's side here, I mean a subscription product also subsidizes a free product. Yes.

That that's exactly right. So, you're going to have tears, right? Free your college, your your kids version free. And those those those versions are going to have more constraints because they know high school kids, you know, sevenyear-old kids are going to use them. So, there'll be all kinds of limits built in. Yeah.

But then you get your $20 version and Jeff GPT is doing what$ 122 billion annualized and growing quickly. Yep. That's insane. Love, you know, lovable. All these guys are just crushing it. And like I I in I was the first investor in a company called Cynthsia which does you know talking avatars.

I don't know if you know these guys. Cool. Yeah. They're killing it. Killing it. It's just up and to the right. Yep. You start them off with it's just a whole premium model, right? That's been around forever and I think that's going to happen. But as you know we talked earlier about buying IP.

You'll see things go into different directions. So you'll have a chat GPT/halthcare and if you want that it's an extra five bucks a month. You'll see chat GPT programming or math or foreign lang dualingual version right and it might be an extra 50 cents a month or a dollar a month whatever it may be.

And then there'll be, you know, entrepreneur version and it'll have all the integrations into all the the states so that you just, you know, fill in these forms and you're incorporated wherever you want to be incorporated because all the agents will do all the work. All that stuff you can upsell. Yeah.

And so, but if they're not using your model in high school, they ain't going to use it in college. And if they ain't using it in college, when they go for that to that company, they're not going to use you. Y and I think that is a big deal. But I can't say it enough.

All these people, we're in a trans transition period right now for kids coming out of college, whether you have a computer science um degree or not. But you're not trying to go to the big companies to get a computer science degree. It's probably not the right way to go.

going to any other company who has no idea about AI but needs it in order to compete. There'll be more jobs than than people for a long long time because there's only going to there's going to be two types of companies in this country in the world.

Those who are great at AI and everybody else and this is not AI is not intuitive. You're not going to take a 40-year-old who's worked at a company for 15 years and say, "Go play with chat GPT and figure out how we can improve our our productivity and improve our processes.

" Some people will put in the time to learn it, but it's not intuitive to pull all those pieces together and extend them with agents and have the agent start doing things and then, you know, writing software on lovable or whatever and having all that's going to be intuitive for for digital natives and Gen Z who are going through school and graduating with that.

That is going to be jobs left and right. For sure. Totally agree. uh wanted to get your we were just on with Vlad from Robin Hood and he is he's super excited about getting uh everyday investors access to great private markets uh great companies in the private markets. I wanted to get your general read on that.

You've obviously been part of Shark Tank for a long time. Can you imagine if you were watching you could be like I want to invest. That seems extremely long. Yeah. And a lot of people have taken crack. I'm sure people have pitched you this exact thing.

But I'm curious because because you obviously you you uh I can I can see you being Yeah, it's great when company when if retail can get into the next Figma, but there's, you know, not making a lot of, right? I' I've been anti- um not sweat equity um crowdfunding, right?

Crowd equity programs against it because there's no liquidity. People love a company, the next Figma, and they put in their life savings because they know Figma is going to the moon, baby. and they can't they can't get out. Yeah. And they got to wait. That's the problem. Y and so that's part one to the problem.

Part two and more of a system systemic problem. There's just not enough IPOs. I mean it's like watching some of these IPOs. They're like meme coins now. Bam. They take off and then it's musical chairs to see who can get out the last but first, right? Yep.

And if you don't get out first or at near the beginning, you're getting crushed, you know. And so is there a way whether it's tokenization or other approaches to that? Probably. But it's going to have to go back to the future where somebody is going to have to take some risk and be a market maker. Yeah.

If Vlad or whoever is a market maker so that there's always some liquidity within X percentage of the last trade, great. then everybody can do it and everybody's protected to a certain extent as long as they're not doing deals where the people are just out and out lying, right?

Or misrepresenting it, at which point hopefully the CFTC or the um SEC, whoever happens to end up in charge. It's not enough just to bring access, you also have to bring liquidity. Uh last question, if Donald Trump runs for a third term, will you run for president?

I mean, it's easy to say yes right now, but it's that's the probably the only thing that could make me really reconsider because otherwise I'm not I don't want to do that It's not my not my dream to be president. I'd rather argue with you guys about my dream when you come back on the show.

You'll be too busy if you're in the White House. Last last question from me. Are you excited about the the the the American taxpayer potentially taking a position in intel? Oh, interesting. You know what? It's a great question, Jordy. So, you're sounding you're sounding like an AI. You're sounding like an AI.

Yeah, you sound like an AI. You know what? That's a great question. What you don't know is I have Chat GPT listening. Yeah. Yeah. Yeah. Yeah. He's running clearly. He's running clearly. It is a progressive dream. This is the Bernie Sanders dream. The AOC dream.

because and I'm not saying I agree with it, but if you think about it, yeah, it's asking for equity and taking value from the billionaires before it even makes them billionaires or goes to the billionaires.

So, when you take 10% of Intel, and first let me add, the fact that it's a grant that they're converting, I think is awful because it was meant to be a grant and that was the deal the United States of America made and now we're going back on the paper that we signed as a country. I think that's bad.

But generally, if this was okay, we'll put in more money in order to take and we want 10% of Intel, obviously, they can say yes or no, just like Nvidia and AMD had the option to say yes or no on the H20 chips. Just like material, whatever MP um had the option. Yeah.

Politician politicians have gotten so good at trading stocks. Maybe we want them maybe we want them running our our federal No, you don't. time. Yeah. But if you're going to generate if I if I was running right and I would say Donald Trump did the same thing, he got this one right now.

The only question is will he do it at the right time? Because would you I rather take more money from taxpayers, the average person who might need to pay more or even the the billionaires, the oligarchs according to Bernie, would I jack up their tax rates to 30 from 39 37 to 39 or 41 or can I just take it this way?

Right. Take it before it even gets to them. That's a progressive dream. I don't understand why AOC and Bernie aren't shouting to the rooftops and Elizabeth Warren. Yes. Yes. Yes.

And make, you know, make Donald Trump's head spin around in circles along with every other supporter because those guys if they got excited about it, he might reverse course. Well, of course, that's why they're being quiet because they don't want him to reverse course. This is 40.

Maybe you're ahead of chess on both sides of the aisle. Everyone's playing chess. But don't you guys agree? Isn't it just like a progressive dream? Yeah.

Like, and you know, so tariffs, which is just like the greatest sales pitch in the history of sales pitches, when you can convince 30 plus percent of people in this country, that that tax that they're paying with increased prices isn't a tax. Somebody else is paying it. Amazing sales job, right?

So now that's $360 billion towards the deficit and everything.

And if you can start doing these other deals that aren't pigish, right, meaning they're they're fair to the companies that are involved and they're not across the board like tariffs and start banging down on the deficit in the way that is the ultimate progressive approach. He's turned into Bernie Sanders. Yep.

He literally has, you know, only a better salesperson. Uh, you're great at this. This is I mean I I give him credit though. I'm not knocking him, right? I am giving Donald Trump all his flowers because he is he is doing things in a way that the Democrats aren't smart enough to figure out to have done themselves. Yep.

This is great. Thank you so much for joining. We'll have to do this again soon. Yeah, we got to do this again soon. We got to find a new uh new topic to argue about, but this was fantastic. Healthare. Let's go. Healthcare. I got Wait till you hear my with healthare. Okay. Yeah, let's do it. Let's do it.

Healthare debate soon. Well, we'll email you. I'll see you soon. Have a good night. Thank you so much for joining. Appreciate it. Talk to you soon. Uh up next, we've been keeping and waiting. We got Quaser Ununas from Clyde. Uh welcome to the studio. Do we have him? I want to ring the gong preemptively.

I don't even know if we