Mark Cuban: AI Hype vs. Reality, OpenAI's Wasting $1 Trillion, Lebron vs. Jordan

追加: 2026-04-29

[00:00:00]We have a great crowd today and we're definitely going to be putting this on YouTube. So, I want everybody at home to hear how many people we have in the crowd. Make some noise.

[00:00:12]I'm Alex Canitz. I'm the host of Big Technology Podcast and I'm thrilled to be here with Mark Cuban for a conversation about everything AI. Mark, welcome.

[00:00:20]>> Thanks for having me. So, we're in the middle of this moment where we're hearing about the revolutionary power of artificial intelligence where every day new capabilities are talked up.

[00:00:30]>> I've heard that. Yeah.

[00:00:32]>> And despite the hype, our day-to-day isn't changed very much. For most people, chat GPT is something they might try one day or it's a better Google. And so, an AI agent is something that, you know, sounds like it's in the future but isn't applicable today. So, how would you describe the gap between the hype and the reality on the ground today?

[00:00:54]>> I don't think there is a gap. I really don't. I think, you know, if you're not using one of the large language models, whether it's Claude, my favorite chatt, Grock, um, Gemini, you know, from a business perspective, since this is a business and student audience, you're falling way behind. that if you don't know what an agent is and you happen to work at a company or run a company, you're falling way behind. Um, it's not that AI is smart. It's not like Arnold's walking through the door. Um, it's not like it's going to take over everything, but like every other technology tool that preceded it, um, it's it's having a major impact. I mean, I've been around long enough. I remember selling PCs to people who said, "We don't need PCs."

[00:01:40]And then talking to people saying, "Well, we're going to connect them together." I started a company to do that. Oh, we don't need them to connect them. I'll just take this floppy disc and I'll just carry it over to this PC.

[00:01:50]Then the internet came along. What's the internet? Right. And we started a company, Audnet, which was the first streaming company, and people thought I was an idiot. They're like, "Wait, you want to be able to use this PC to listen to sports and radio and watch things?

[00:02:07]Dude, I'll just turn on the TV and radio. You're a you know?" And then it just goes on and on. And there was always a group of people that were first and always a group of people that were naysayers. And the people that were first typically ended up getting further ahead. And I think it's the same with AI today. But hold on because I was doing my research.

[00:02:28]>> Uhhuh.

[00:02:29]>> Not as long ago as February 2026, you referred to AI as the equivalent of a hungover intern, >> right? But that's agents, right? Because what I was saying is you know what an agent can do when when you put together an agent whether you're using cloudbot you know cloud whatever it may be it does all the the tedious work that you don't want to do and all the stuff that you hope to get to at some point but you just don't have the time to get to. So you get that intern probably hung over Friday morning and that's >> is there any other kind?

[00:03:07]>> Right. And so, um, it's easy to get all those things done using an agent. And, you know, you can let the intern stay out later, come to work later, you know, because this agent's going to work 24/7.

[00:03:20]>> Now, you've expressed some skepticism along the way. I definitely agree with you. Right now, we're in this moment where you need to be using it, but I think you've been measured in your comments around AI in a way that some of the hype folks have not. And so, here's the question to you. you've been on the front lines of a couple of big changes in technology. The way that the AI labs describe this moment is that it is an exponential, right? It's it's kind of an amazing thing where the horizon seems to be getting closer as opposed to further away. And they say that makes it different than our typical technology shifts which tend to progress linearly.

[00:03:55]They're important but linear. Where do you fall? Exponential or linear?

[00:03:59]>> Exponential.

[00:04:00]>> What makes you say that?

[00:04:01]>> Just because it's changing so rapidly.

[00:04:02]you know, when it's linear, like back in the PC days or even your the first iPhones, remember, you know, you got your first iPhone and then you heard the news about a new iPhone and that iPhone was faster, more memory, you know, better connectivity and that's linear, right? You know, just one step and you wait for the next step and then you get to a point of diminishing returns where like you don't really care whatever number iPhone we're at, right? It's good, but there's nothing so stupendous that's going to make you have to replace. With AI, the changes are just coming so fast. You know, the agentic AI, the ability to, you know, do so many different things. If you're trying to program, being able for anybody like, you know, for cost plus drugs, I like to compare our pricing to other competitors. And I just went into cloud and said, "Okay, go to these three sites. Here's a list of all the um the products that cost plus drugs carry.

[00:04:57]Take the top 25 most expensive and every week I want a report comparing the prices.

[00:05:04]12 minutes later, boom. I mean, that was inconceivable. You know, I I can write software. At least I could write software, you know, or you can hire somebody. You can even get a drunk intern to do it. But to be able to tell Claude on your phone to do it and say, I want to be able to click on it from any browser and, you know, it's going to have a couple issues. No, this isn't right. Reformat this. I'd rather have it this way. And then it just goes back and forth. I mean, that's nonlinear. That's up and to the right.

[00:05:32]>> Uh, tell me about the origin of this drunk intern metaphor. Have you encountered many of them? Were you one?

[00:05:40]>> Where's my beer?

[00:05:43]>> That's after this. I You know what?

[00:05:46]There's not a lot of jobs I've excelled at. That was one of them.

[00:05:53]If this is an exponential, then where does it go? I mean, with the iPhone, you're right, it plateaued. iPhone 15, 16, 17, they all look the same. I was speaking about the iPhone earlier today, and I couldn't really even mention what number it was.

[00:06:06]>> I I couldn't remember either just here as we were talking about it. Yeah.

[00:06:09]>> And it's my job to follow this. But if AI is not going to have that plate >> a drunk intern or claude >> or clad.

[00:06:16]>> Um, so if AI is not going to have that plateau moment, where does this moment of technology go?

[00:06:23]>> Well, to say it won't have a plateau moment, I that's different than it being up and to the right and exponential.

[00:06:29]Okay. Right. it can continue to grow exponentially and have and change its features and change what it can do and change how it can change business and education and so many different industries and then reach a plateau because like right now if you just showed if you I I'll give you the perfect example where AI isn't so good right so if you were blind and you had a um seeing eye dog and you were at an intersection and you also had a cell phone with AI and you had the choice between using your seeing eye dog or holding up the phone and having it guide you. I'm taking the seeing eye dog every time. Right? When you think about AI today and for the foreseeable future, it's not going to know the um it's not going to know the consequences of its actions. Right? So if you take if you ask Chat GPT or any of them something and it gives you bad advice, it has no idea whether you know what's going to happen because you took that bad advice. But a 2-year-old that is on a high chair with a sippy cup, he knows or she knows that when she pushes that sippy cup off the the high chair, mom's going to come running and the baby's going to be laughing its ass off, right?

[00:07:52]it knows the outcome of its um recommendation or if it's it knows the consequences of its actions and that's a big difference right and so while AI can help as a business tool it's not the end all beall where it's going to make all decisions for you and I think right now we're bifurcating into two types of ways or two types of people that use AI people who use AI so they don't have to learn anything and people who use AI so they can learn everything. And you see the difference there, right? I can go to AI and like, okay, I just got to come up with this report from my boss. Here it is. I put it in. It gives me the research. I give them a report. Okay.

[00:08:33]Or, oh my god, my boss is interested in this subject. They want a report. I want to learn everything I can and I want to be able to use AI, you know, different AIs so that I can come up with the best um resolution for what they're looking to accomplish. I know I want to learn more about all this so I can be better prepared to help them in the future. And so people get nervous when people just use it, you know, just w without without caring or trying to learn. But the opportunity to to use it to learn, it is the great democratizer of knowledge like we've never seen before. I mean, you can be a 8-year-old in the worst set of circumstances in your home, in the worst neighborhood, anywhere in the world, but if you have access to a smartphone and you can go to chatgpt.com or whatever it is, or claude.ai, you have access to every library, pretty much every professor, every consultant, and you can ask any question. Oh, okay.

[00:09:31]I'm interested in learning about social studies, the constitution, right?

[00:09:38]whatever and you can just say teach me and it'll put together a program to teach you. And so those people who are using AI to learn, those people who are curious and just want to keep on learning more, AI is phenomenal. You will always have an edge over everybody around you. If you're using AI to learn, if you're just using it just so you don't have to do the work and it's your drunk intern, you I mean, you're going to struggle.

[00:10:07]One quick tangent. Actually, the way that AI gets better is very much like that child with the sippy cup. The way that these labs have been training these models is you have a desired behavior.

[00:10:17]When it gets there, you reward it. When it doesn't get there, >> you don't punish it, but you don't give it the reward and it learns to optimize.

[00:10:23]So maybe it can even press, >> but you have to define the optimization that you're looking for. You have to define all the goals. You have to know all the context, right? In real life, we don't know the context for everything.

[00:10:35]We don't know what's coming next. We don't know how soon it's going to happen. So, not all that information is available to train those models. It's like if you just went to any AI right now and just said, "Tell me how the war in Iran's going to turn out.

[00:10:51]Whatever it says is, you know, I mean, pretty much useless."

[00:10:56]And that's the way things, you know, so you have to know where the output of the AI you're using matches what you're trying to accomplish. and just training on like for games it's easy right because there's only so many permutations but for real life trying to figure out the context it's much harder and it's not nearly as as valuable but within business we tend to have goals we tend to have standards we tend to have a path we have processes and AI is a great tool to allow you to optimize all those things >> so how do you build a business today given the tools at your disposal is it different than the way you would build a business previously I mean you're trying to serve customers. Do I think is do I turn over my customer service to these GPT models? But I imagine there's more fundamental questions about the way to build a business.

[00:11:48]>> Well, it just depends where you are in your business life cycle or are you starting a business? If you already have a business, the things that you've considered outsourcing, you know, your um customer center, your customer call centers, whatever it may be, AI agents are perfect for. So, you can do that.

[00:12:02]the stuff you would give to an intern, AI agents are perfect for. Um, but if you have a big business, let's just, you know, we hear a lot of stories about huge public companies that they try, they spend a lot of money to try to implement AI and it doesn't really get them a return on investment. And people say, why if AI is so great? Because you already are running your business the way you've always run it. And in order to take advantage of AI, to answer your question, you have to reformulate your business completely to build it on AI.

[00:12:35]It's the difference between running a business before PCs and computers and running a business after PCs and computers. You would do it completely differently. So starting a company now using AI, I mean there's so many different ways that you can use, you know, Aentic AI to do processes and optimize um steps that you would take to communicate um and then just to use the large language models to you know to learn and and learning is not the right word but to accelerate things right using you know a chat GPT or collide is a great typing hack like we got a a letter an email today from somebody asking us something about our website and there was a legal question associated with it. So I just cut and paste the whole thing into Claude and I put a link to the website and I said show me all my options that I should should consider here. You know there's just those are things normally in the past I'd be like okay I got to get the lawyer I got to get the person in charge of this group and that group. Now I just run it through, learn it, understand it myself. Then I go to that person and say, "Is this correct?" And I spend far less with the lawyer. I spend less time optimizing with the person ahead of this group, etc. So when you're working to create a business, you're going to take a completely different path using AI than you would preAI.

[00:13:57]>> And and that's why, you know, to go back to the question I led with, we're why I think we're seeing this moment where we have this technology. It's very powerful, but it's not permeated through everything we do. It's because what you just described, that process you described of there's a question. I'm going to go to Claude and maybe not a lawyer and work through my options. It's great if you're at the top of the company. In a company that, as you mentioned, has done things their way for a very long time. There's a certain granting of entrepreneurialness that needs to happen for somebody who's not at the top to be able to go in and say, "I'm going to actually use this technology to its potential. I'm going to be in entrepreneurial within established company."

[00:14:42]>> Absolutely. You know, like I told my daughter who's getting ready to graduate from um college and get a job at a consulting company, you have to learn this because if you're not the person who knows, you know, how to do vibe coding or do all these different things with agents and claude, somebody whoever does is going to take your place.

[00:15:00]>> But isn't the bigger question whether her boss will allow her to do that?

[00:15:03]>> No, I mean it isn't. It isn't, right?

[00:15:05]You have to know, you have to use this as a tool and understand how it works.

[00:15:09]um so that as it evolves over time, it can change and changes rapidly. You can keep up with it. If your boss enables you to use that extra knowledge, great.

[00:15:18]If they don't enable you to use that extra knowledge, extra knowledge, they're not going to be your boss very long. And if the CEO doesn't understand that, he's not or she's not going to be the CEO very long. And if they still keep that CEO who's not using AI to get ahead, you tell me so I can start a company to kick their ass.

[00:15:39]Because because that's reality, right?

[00:15:47]Literally, you know, when we started streaming and I gave you the example, it was like, okay, it's a PC now and it's a pain to to be able to stream, but someday you'll just click and the video will come on. And that's what we have today. you know, you have to be able to see where your business could go and understand how to use the tools available and, you know, kind of guess where those tools may take you. And particularly now with AI, going back to what I said earlier, the biggest challenge for a company is going to be for the CEO to make that decision that, hey, we're going to have to blow up a lot of what we do to recreate our company because if we don't, somebody else will, and they'll have far less expense to be able to do it. When it's all said and done over the next three years, there's going to be two types of companies. those who are great at AI and those who went out of business.

[00:16:42]>> Yeah, we could see a new era of competitiveness because >> there already is.

[00:16:46]>> The door is going to be open to those to come in and and um and disrupt.

[00:16:50]>> It's already wide open. Like, you know, all of you could take out your phone right now, go to your favorite large language model, come up with an idea for a business, instead of coming to me and saying, "Mark, is this a good idea?" Um, just go in and said, "Okay, here's what I want to do. Here's what I think. write me a business plan and tell me the good and the bad and the ugly about this, how I should finance it, what the economics should be. 10 minutes later, I I'll give you an example. When I was trying to teach myself how to learn how to use this stuff, I'm like, "Okay, I want to come up with an idea and see if I could take it all the way to a patent." I'm like, "Okay, so let's imagine that we have a button that we'll put on a shirt that has a video camera in it that can record 24 hours, then automatically send it to my phone or my PC, everything that it's recorded, and save that for me. And I want a bill of m I want a business plan and a bill of materials and a patent application 12 minutes."

[00:17:43]Now, was it all absolutely correct? No.

[00:17:46]But I went back and said, "Okay, get me vendors and approximate cost for all the bill of materials." you know, everything that goes into a product, minutes. And not that it had to be perfect, but think about what had to be done before. If I wanted to go through the same process, I'd go to Google, hey, I'm thinking about putting this together, da da da da da, and then you see 15 blue links, and seven of them would be garbage, like you just wasted your time on. And then you'd have to go through all of them. Now it it you know composits it and puts it together in the way and the format that you want based off of everything it's been trained on and you know the probability that this is the best response and any questions you have you just ask it. Just the idea that this is a time hack, right? Just a time-saving tool that allows you to fill your curiosity and learn things that's insanely valuable for anybody everywhere. Like I said, you could be a six-year-old, you can be a 96 year old trying to figure something out, and this is just going to allow you to do it more quickly. And as it gets more powerful, it's just going to have more impact.

[00:18:52]>> But on the other side of that, like, isn't it a little bit annoying to you as an investor because of the slop coming in? You're someone who famously answers your email. I'm sure you've noticed something in your inbox that many of us have noticed in our inbox, which is that a lot of inbound emails now sound exactly the same because written by Chad GPT.

[00:19:11]>> But I created an agent to look for those things that take them. Seriously. Um, you know, you get a lot of garbage, but I always got a lot of garbage. Like, you know, I on Shark Tank it was like people would have the next Uber, the next this, the next that, right? Because people just copy. And I I've always said, you know, if there's a thousand people doing the same thing, why do you want to be the thousand first? Or if there's 10,000, why do you want to be the 10,000 the first? And it's going to be the same thing with AI. Now, is there going to be kind of a a revolt against AI at some level? Yeah, of course. You know, we're going to do more things in person because there is so much in AI. You're going to be, you know, more interested in face-to-face communications, face-toface events than you are going to be in, you know, the 10,000th email that said, "Hey, I saw your LinkedIn profile and you look really smart and I think you'd be great for this job." I'm like, "You're sending that to me? Seriously?

[00:20:10]>> Did you take it?"

[00:20:11]>> Um, I applied.

[00:20:16]Um, but you know, so yeah, there will be things that are just over the top, but that's because we're in the beginning.

[00:20:23]Everybody's trying different things because look, you know, a lot of that is going to go right into our spam folder.

[00:20:29]And, you know, just like we get spam calls still that are just annoying as hell, right? But at least it tells you now it's spam whereas before it didn't.

[00:20:37]And the same thing will happen with your email inbox and you know other communication tools where AI will advance far enough where it'll be a little battle between their AI trying to get through the spam bots or the spam filters and you know that that's just part of the progression. I mean, isn't it a cool part of this though that while it does definitely democratize the ability for people to pitch an email, it also creates the ability for people who may not be technical to build that minimum viable product and actually show it to you.

[00:21:05]>> Oh, yeah. For sure.

[00:21:06]>> For all the people that I'm sure you've heard and we've all heard, hey, I have an idea for an app. Do you have a developer? No.

[00:21:12]>> No, >> you're never going to do it. But now you're really able to do it because you can have cloud build%. And that's probably >> even worse, right? It's just like, hey, I need to raise some money so I can go out and get a developer.

[00:21:23]>> Not anymore.

[00:21:23]>> Not anymore. Right. And if you're not willing or don't realize that you could just go to Claude and learn this stuff and do an MV a minimum viable product, then you're not somebody I'm going to invest in because the number one um skill set that you can have as an entrepreneur or as an employee or as a student is curiosity.

[00:21:44]And there's no better way to fill that curiosity and keep you curious than AI.

[00:21:49]I mean, when I was younger, I used to go into bookstores and sit in the aisles and look at the books I couldn't afford to buy or go to a library. The same thing. Or then, you know, like many of us, when the internet hit, it'd be like, okay, and you find yourself going down these rabbit holes on the internet. Um, but now it's just so much faster, easier, more direct. um it it's just a better tool that you need to learn how to use.

[00:22:13]>> There's one more side to this which is that the market has been really punishing software stocks because it's become that easy to build software. And I think one of the things investors have realized is that all this software that's being built is built to scale but never perfectly suits people's needs.

[00:22:32]There's a lot of room to disrupt them.

[00:22:33]with AI potentially you can have a lot more custom software and they're calling it the SAS apocalypse.

[00:22:39]>> Yeah.

[00:22:40]>> Do you think that that selloff is justified?

[00:22:43]>> Yes or no? Depends on the the company.

[00:22:45]Right.

[00:22:45]>> So that's great because I'd love to know who is safe and who's going to be disrupted.

[00:22:53]>> It's uh Yeah. Anyways, I won't go there.

[00:22:56]Um >> how about how about just the fundamentals of these type of companies?

[00:23:00]>> Yeah. What you got to have is some unique IP that is not shared.

[00:23:05]>> Okay, >> so I'll give you an example, a company called Docu Sign, right? All the things that we sign, you would think that's easy because there's other companies, Adobe and others who do all that. But it's a global company and all the laws associated with signing a legal document are different from city to city to state to state to country to country and you have to know all those things. So that's a unique differentiation that you don't want to be dependent on the AI having rights. So that's an example. Um >> the push back on that quickly is just that the AI agent will go and find the local regulation input.

[00:23:39]>> No, but you've got to go you got to know where to send it to go first and what are you going to have it do um spider every single municipal municipality in every city in every state in every country. It's just not it'll cost you more in tokens than it will in value, right? Than you'll get in value. So I think if a company has IP that as part of their software, when I say IP, not just okay, our software is a little better, faster, cheaper, more this is a database of things that you can't get anywhere else that they're not making available to um to AI to be trained on.

[00:24:17]It's their classified information. um then I think those companies will be okay. If they just, you know, if they're just sold by the seat and it's just a standardized piece of software, they're in deep dudu.

[00:24:29]>> Yeah. When Anthropic, uh there was a rumor that Anthropic was going to release a legal plugin.

[00:24:35]>> Mhm.

[00:24:36]>> The entire legal section of the software just went poof.

[00:24:40]>> Yep.

[00:24:40]>> But that that does seem like though maybe that's ill advised because they do have proprietary information. Well, yes or no because all those are coming from Pacer and other legal databases. Right.

[00:24:49]>> Okay.

[00:24:49]>> Right. Because I I talked to a law firm because I was having the same type of conversation and I'm like, can we take all of your, you know, legal briefs and this and that and train from them? Like, no, they're proprietary, A, and B, they're specific to a client and we can't share them. And so, but that's for a law firm that's not for the legal software.

[00:25:09]>> Okay. Speaking of another area that might be disrupted, uh, consulting is a field that people talk about often as things like implementation and doing business analysis. You can prompt that and one-shot it probably. You mentioned that your daughter is going to work in consulting.

[00:25:24]>> Mhm.

[00:25:25]>> Did you bless that?

[00:25:26]>> Yeah, of course. Yeah, because Yeah, go ahead. there going to be a period of time where somebody's got to help all these companies figure out how to implement AI um and not only implement it but how to reformulate their businesses because how you do things in an AI universe where you're trying to be more productive, more profitable, more competitive is going to be completely different. The hardest challenge for those CEOs is am I willing to blow up my business um knowing that my stock price could collapse in the meantime and then I have to build it up and working for a consulting firm you know you hopefully have a lot of proprietary information that will allow you to guide that CEO as they go through that process.

[00:26:10]>> Let's talk briefly about the business of AI. Mhm.

[00:26:13]>> I mean for from someone in your perspective looking at the amount of investment that's gone on to these AI companies must be fascinating. OpenAI recently announced that they had raised $110 billion which is I think 3x largest IPO ever. They're planning to spend more than a trillion dollars in infrastructure in the coming years. I'd love to hear your perspective not on the spend but on the return needed to make those numbers work.

[00:26:39]>> They'll never get it.

[00:26:41]>> Yeah. They're they're they're just away that money.

[00:26:46]I mean, at scale, right? It's not that AI is not going to work, but look at Apple, right? They haven't spent next to anything, but they've got a foundation where they can just plug and play into their devices. But there's a lot of FUD being put out about the spending, right?

[00:27:02]We're going to spend a trillion dollars because we need all this data center capacity. Well, the data center, the the ability to process is going to get faster, cheaper, right? And it's going to happen faster and quicker than people expect. And so, I think a lot of the numbers that they're throwing out there um aren't going to come to fruition, you know, which is why I say they're full of It's not going to happen. But those who have just gone allin, some of them are spending more cash than they have available. And I I understand why.

[00:27:35]I just don't I'm not convinced that for all of them it's going to work. And the reason why is we don't know if the the business of foundational models the chat pts geminis um grock um claude etc is going to be like the streaming industry where there's one leader and a bunch of players that make money um or some that make money at least or search where there's effectively one company and so they need to raise all the money go all in and spend everything they can kiss all the rings They need to kiss around the world in hopes of being the one, right? It sounds like um Morpheus, I'm just waiting for Kiana Reeves to walk through the door.

[00:28:18]But so it's it's going to be very difficult, but I don't think it's going to pay off the way they expect.

[00:28:23]>> They're away the money at scale.

[00:28:25]>> It's hard to do, right? But but if you don't go all in like that, you can't keep on raising money because as it zigs and zags and you don't know where it's all going to go, if you're not still raising money, you're in deeper trouble because you don't know if you know if you're not the winner, then you've got problems. Let me make their case for the sake of argument. Um, right now what we're seeing is a moment in the AI world where the training has gotten to the point where it's less general.

[00:28:56]you focus on certain capabilities. We know anthropic has focused on code, open AI has focused on health care and it's difficult to train up really well in those disciplines and you see things like anthropic leading in code for a while because they've gotten good at it.

[00:29:12]So that to me would suggest that there's going to be multiple winners here potentially. But in a vertical, right, like healthcare, you've got open evidence and open evidence is actually going out and buying intellectual property.

[00:29:27]Um, chat GPT is not. And what you'll you're starting to see happen and will happen even more. If you own intellectual property, you're going to see fewer people applying for patents because the minute you apply for the patent, it's available to everybody, which means every model can train on it.

[00:29:43]And when I go and talk via cost drugs, when I go talk to hospitals and research um um campuses and um schools, I tell them, "Do not publish, sell." Because the minute you publish, even though it makes you feel better and you seem like you're, you know, a bigger dog in the community, um all of a sudden every model has been trained on it and you've given away your advantage. And so when it comes to verticals, I don't know necessarily that they're going to be able to buy the IP they need going forward. Now what um Claude is doing with Anthropic and focusing on programming that's not really a vertical per se, right? Because it can be used for everything. It's it's really process driven and I think that's smart. Um but when I look at the other ones, you know, Gemini has an advantage because Google search is Gemini, right?

[00:30:37]And so now when you do a search, it comes back with an AI response and it can sell ads around it. I don't know what Meta is going to do. I don't know what um Open AI is going to do when it's all said and done because I don't think you know you can I always call like healthc care for open AI is a feature.

[00:30:56]It's not a product and it's a feature that you know open evidence has that others can add just by going spending money on the IP. And then you've got the bigger problem like when you think about all these foundational models today and for the next few years, they are built on text and pictures, not really video.

[00:31:15]They don't know what's happening in the real world. You know, like I mentioned the the example about the 2-year-old pushing the sippy cup off the high chair. Like it doesn't understand that.

[00:31:25]you know, it's not if if you fed it all kinds of information, it's not going to be able to come up with E= MC². And to do that, you have to understand physics.

[00:31:35]You have to understand the world. And so, we're going from large language models, I think, to what's called a worldview approach where it's built on, you know, understanding the physics of the world and being able to take video and understand it. I invested in one company that's putting up satellites that can look at different materials and tell you what the makeup of those, you know, um, natural materials are. Is it wood? Is it rock? Whatever is in there, like a spectrograph. And I think that's going to supersede a lot of what Claude and Grock, etc. are doing. And so, we haven't seen the best of AI in my opinion. Um, and it's going to just get crazier and crazier and crazier. And so when you think about all these people spending all this money, look, if you're the best programming, you know, like Claude is, and that's your niche, great.

[00:32:25]But I don't know what the niche of these other ones are. Do you?

[00:32:29]>> No. But I I don't think there's going to I mean, if you have three, four winners, maybe that works. Maybe three. Any more than that, it's going to be difficult.

[00:32:38]>> I mean, it's like then you're then you're just a website.

[00:32:40]>> Yep.

[00:32:41]>> You know, you're just an app. and you just spent a trillion dollars to be an app.

[00:32:48]It's a lot of money.

[00:32:50]Let let me ask let me ask you this. You're someone who gained national notoriety by being, you know, somewhat brash and rebelling against the establishment.

[00:33:05]that. No way.

[00:33:09]Uh, do you see some of yourself in the Sam Alman's and Dario ammo days in the world?

[00:33:15]>> Maybe a little bit in Dario, but not in Sam. Um, say more.

[00:33:20]>> I mean, you know, Dario now is trying to scare the out of everybody, right?

[00:33:24]But that's part of raising money. Like when we had Broadcast.com back in the day, I would say, "We're just going to replace cable and satellite and all that." And it didn't, you know, but I kept on saying, "It's going to be in a couple years." And here we are 30 years later and we've kind of gotten there.

[00:33:39]And so I think Dario is pushing forward with Claude and I think to your point their niche with programming and agents I think is going to hold for them. Um and I think they're they understand they may have to move to a worldview environment as well. Sam is all over the map and I think that'll backfire on him.

[00:34:00]Um, I just think, you know, they were they were buying up 40% of this company's memory chips, right? And they just backed out of that. You can't do that. You know, at some point people stopped trusting and stopped wanting to do business. You know, they just moved him to from their safety committee, I think, to something else. I forget all the details. I think, you know, Sam's accomplished a a bunch of amazing things, but I just I just don't see how either one of them communicating is beneficial for anything but raising money.

[00:34:33]>> So, what would your advice to them be?

[00:34:35]>> Um, they're doing they don't need my advice at all. Right. Um, you know, you just it's so early. It's just changing so fast. You you know, you've just got to stay connected to as many people as possible. Um, but I think you have to be more involved at lower levels as opposed to just programmers. I think, you know, Apple's got it right in some respect in that everybody with an iPhone, it's going to be plugandplay to plug in whichever model you want and they'll have the ability at some point just to swap those out for their own if they choose or just like they've done with search engines, just pay or be paid. Um, but I think you want to be where kids are and showing schools how to use AI so that it's a democratizer for education and a tool for teachers to be able to, you know, bring out that light in every student's eyes. You know, right now when we teach, we teach by the stu the teachers there. They create a syllabus.

[00:35:39]They've got the subject matter and then we teach everybody the same way and then we expect all the students to know the answers to the same questions and teachers freak out. Well, all you got to do is use AI. Well, the only reason teachers are getting it wrong is because there's they're asking the same questions of every student in the class, right? They're telling them to respond to it with an essay and every student has to respond the same essay. when in reality when I talk to teachers what I tell them is you know the hard part for you is to get that light in the kids's eyes to get them really excited about a subject matter and with AI you can personalize teaching to each individual student you can say okay we're we're doing um American history and we're learning about the constitution you know little Billy um in sixth grade or whatever is maybe isn't as good a student so I'm going to we're going to put together um a project under AI that works at Billy's speed, reformulates the questions for Billy based off of his answers, and then keeps the teacher and parents in the loop so they can see where little little Billy is. And hopefully you can even mod, you know, you can even have the AI personalize it so that it's more interesting. You know, if they like learning more about Paul Rivere than George Washington, right, you spend more time on Paul Rivere to get them more interested. You can personalize any topic with AI. You can't do it with traditional teaching. And going back to your question about, you know, for the AI companies, I think by getting students and teachers excited about the capabilities and features of the AI that you're offering creates a foundation that that they'll they'll want to be comfortable with and they'll stick with those tools for a long long time. Isn't it true that personalized tutoring to a student gets you like two standard deviations above the mean?

[00:37:34]>> I I don't know. I don't know. It just makes sense to me.

[00:37:37]>> Let's Someone knows. No, you're just laughing. Okay. It's true. From the front from the from the front row. I'm pretty sure that's the numbers. And it's just been inaccessible to so many people because of the expense. But if you're able to scale that up, >> there's time there's expense. And you know, there's a difference between homework. teacher said, "I got to write an essay on da da da. I've got to memorize da da da da da." And going home and on your laptop or phone or whatever you have available to you, you know, having it presented to you with graphics and videos and in a manner that's fun that the AI will understand if you're engaged with it or not. And think about that. It's not hard for an AI to know if you're answering the questions. It's not hard to know if you're g giving more engaged answers. It's not hard to know if if you're asking it question or if that if the student's asking questions, right? And it's not hard to grade the test and grade the answers and then summarize it for the teacher and the parent and then the parent can make suggestions based off of conversations.

[00:38:40]The teacher can. That's just so unique.

[00:38:42]And not only does it apply to teaching kids, it applies to teaching anybody anything.

[00:38:49]Like if you've ever used Notebook LM from Google where you could take it and turn whatever you want a training manual into a podcast.

[00:38:59]I mean it is insane. You know, doing a training podcast for, you know, how you want to keep the restrooms at the American Airlines Center um clean. And it's a man and a woman. Yeah. We were just talking. We were just going through the training manual of how to keep the restrooms clean. This is really interesting. The the tools that they use. I mean, who ever thought that this mop could work like this? And they made it so, you know, cuz kids particularly today don't want to read. But if you put it in a podcast format, particularly one that makes you crack up when you're talking about mops and, you know, how you're going to mix the solution to clean the floors, it's just changed how we educate people. And if you take advantage of that and learn from it and learn to apply it, you will have an advantage as an entrepreneur, as a CEO, as a manager, as as a an employee, as a student, as a teacher, all across the board. If you don't, and you're doing it the same old way, you get the same old back.

[00:40:00]Who Who does succeed then in this era?

[00:40:05]Whoever learns how to use the tools the best, you know, whoever realizes that the operative word, the key word in an AI world is iterate.

[00:40:15]As the tools change, you have to keep on iterating, iterating continuously, recognizing you're going to get smarter, your employees are getting smarter, your customers are getting smarter, agents will communicate, and they'll be smarter. So you have to be in a position, you know, as an entrepreneur.

[00:40:30]I know what it's like to, you know, have thresholds that we all try to hit. I got a half a million in sales. Yeah. I get a million in sales. Yeah. Making money.

[00:40:40]Yeah. Up to 10 employees. Yeah. You know, and then you start thinking, okay, I've got this company down. My customers like me. But in an AI universe, you have you more so than now, you're going to have to continuously improve because anybody can come in and say they're not improving. They don't understand this new tool and so I'm going to go compete with them. You have to continuously iterate. I've heard from more than a few people that they believe that using these tools inside their companies effectively means they're training the AI to replace them. That has to that has to be true in some cases. What do you think happens in the broader job picture here when this technology continues to ramp?

[00:41:22]>> There's going to be displacement.

[00:41:23]There's no question about it. But I'm not one of those people that says jobs will be optional like Elon Musk said or um 50% unemployment. No. Um there's things that if all you're doing is reformatting, you know, or you're answering a question yes or no, then you know you're there's a good chance you're going to be replaced by AI. But if you're a critical thinker, you're there's always going to be a need for you. If you're a critical thinker and you understand how to use the tools so that you understand where best to use an agent, where best not to use an agent, where best to do, you know, vibe coding, where best not to, and you under, it's like I was telling my daughter, if you learn how to use these tools, and you know how to think critically, you're curious, so you're always learning, you're always going to have a job because, as I said, AI doesn't know the consequences of its action.

[00:42:18]You don't want to be in a business where you just trust the AI and it's like, "Oh what did it do, you know? You need somebody as the buffer to understand how, where, when, and why to or why not to use it." And that's critical. And that's middle management in a lot of cases. And on top of that, you're going to have companies, like I said earlier, that have to completely reinvent themselves. And when you reinvent themselves, you're going to need people that understand how to make all these pieces work. And so I, you know, I think there's going to be plenty of jobs. And if I'm coming out of school, you know, my daughter already has a job, but if if I'm 16 years old, if I was graduating today, or if I was a 16-year-old looking for a job, I would learn everything there is to know about AI and I would go to small and medium-sized businesses and say, "Let me walk in the door." and all those things that you know you need to do that are the bottom of your to-do list. Let me show you how to use an agent to do those things because in doing so, you'll be more productive, more competitive, you'll make more money. And I'll give you a perfect example. One of my Shark Tank companies is called Rebel Cheeseche.com. They sell vegan cheese and they ship it all over the world. And they like a lot of companies that ship products in boxes. You know, for those of you who ship stuff in boxes, you know there's different size boxes for UPS, DHL, whatever, and you start in one zone and there's a different price to ship to a different zone. And then you also probably know that they don't ever charge you the right price.

[00:43:50]>> Does that sound familiar?

[00:43:51]>> Yep. So, what this company did, Rebel Cheese, was they wrote a little agent that took a picture of the box when it's getting ready to ship and determined the size, looked at the price list, look at the invo, took a picture of the invoice, compared them, and when they were different, created the credit request and are saving $50,000 a month. And it's all automated. They don't have to touch it.

[00:44:16]That's where the agents come in and come in handy. And you have to know how to implement that. And most small to medium-sized businesses don't understand that yet. And they don't have the resources to just, you know, allocate people. They don't have the drunk interns or the sober interns, you know, to just go do it. And anybody here who's an entrepreneur knows there's always 10 things that you say you're going to do on Saturday and never get to, right?

[00:44:43]Always. And so that's what agents are good for. And so if I were graduating today, I'd be going to these small medium-sized companies, charging them a hundred bucks an hour. And then because these these AI systems, they kind of drift each time there's a new version, those agents might need to be modified.

[00:45:01]You have recurring income coming in all the time to go back and check and keep them up to date. There are so many kids that took one computer class, you know, know some Python that are going to be able to go do these things there.

[00:45:14]there's not going to be any shortage of jobs there. But if you try to work in a big company and become an entry- level software engineer or programmer, that's just not going to be there because the people in that big company, they're going to have training. They're going to have older people that are in the IT department that know they need to learn this stuff and those those entry- level jobs aren't going to be there.

[00:45:36]>> Don't worry, they're not hiring for those anymore.

[00:45:38]>> No, they're not. Yeah, they're not.

[00:45:40]>> Yeah. Uh you just to to clarify I was thinking that your your advice was the young person should go in-house at a company >> at a small company >> but but why wouldn't they then just start their own business and sell that as a service? Well, when I say go to small medium-sized businesses, that's what I'm saying, right? Go to them and, you know, show them just like the rebel cheese example. Show them all the the things that are too tedious, timeconuming, and you don't know if you'll get a return. Um, if you hire somebody to do it, that's a perfect thing to bring somebody in for a few hours, a few weeks, and have them define agents and allow the agents to do the work for you. And if you don't know what an agent is, you're going to have to learn. And the best way to learn is go to chatpt or whatever and just type in there, I'm new to all this AI stuff.

[00:46:30]What the hell is an agent? Can you teach me how to do it and how it would impact my business? And let me just tell you part two to that. If you as a CEO, manager, employee are afraid to have a heart-to-he heart convers conversation with your AI, you're going to fall behind. And I'm not talking about your personal life. Is this the right guy for me? Should I date this girl? No. But what I'm saying is it's weird to start talking to your AI and then having a conversation and then thinking, "Okay, I need to have another one." Like, I have mine hooked up in my car and I'll be driving and talking to it and I'll catch myself. This is strange.

[00:47:13]But you've got to get over that hesitancy because like some of the answers like I gave you the patent thing, they're good answers. like you were like that is good and if you don't have if you don't do that if you don't allow yourself to feel comfortable with that it your your career your business life is going to be a lot more difficult >> if you speak with it a lot Mark do you ever ask it can you tell me my deepest darkest secrets or psychoanalyze me >> I can be weird giving it that prompt >> um they're they're counting us down couple more minutes Sure.

[00:47:50]>> Do you guys want a couple more minutes?

[00:47:55]>> All right. Uh, good news. I have a lightning round.

[00:47:58]>> Ah, that's never good news for me.

[00:48:00]>> Um, would Jordan or LeBron win one-on-one?

[00:48:06]>> Jordan.

[00:48:07]>> Jordan.

[00:48:08]>> Yeah. No question.

[00:48:12]>> Could Kobe beat either of them?

[00:48:14]>> Yeah. I mean, they're all so good in any given game, one can beat the other. Uh, Indiana alum.

[00:48:20]>> Yep.

[00:48:21]>> Is all right.

[00:48:23]>> Who's yours?

[00:48:25]>> Uh, is Fernando Mendoza going to be a top 10 quarterback in the NFL? Of course. Okay.

[00:48:32]Um, favorite candidate in 2028?

[00:48:41]>> They're saying Mark Cuban.

[00:48:44]>> Bart Simpson.

[00:48:47]>> I don't know.

[00:48:48]Um, I thought we might get that answer.

[00:48:50]He said Bart Simpson. Is there anyone you wouldn't vote for?

[00:48:55]>> Um, yeah, I got a lot of buddies that no chance I would vote for. No, I'm It's not 2028 yet. And look, there's nothing I can say about politics that's going to change people's minds. And so, you know, I'm in the healthcare business with cost plus drugs. Republicans need low price drugs. Democrats need low price drugs.

[00:49:11]That's all I care about.

[00:49:19]That's a winning message. Um, should should the NBA shorten games?

[00:49:24]>> Yes. To 40 minutes.

[00:49:26]>> Say more about that.

[00:49:27]>> Um, anybody here watching March Madness?

[00:49:32]>> Now, if you're a Duke fan, you were upset there weren't eight more minutes in their last game, but nobody's upset that it's 40 minutes. If you're into the WNBA, you're not upset that it's 40 minutes. If you watch Olympic B basketball, you're not upset that it's 40 minutes, you know, high school basketball, you know, all college games.

[00:49:49]And so by cutting back to 40 minutes from 48, it's equivalent of shaving about 14 games off the season, which I think will help um the players bodies.

[00:50:00]And I think it would make the games better because, you know, we all kind of pitterpatter till the fourth quarter and now the fourth quarter comes quicker and every possession um impacts more. Plus, there's more upsets because, you know, there's fewer possessions and every NBA team is still good. And so one team gets hot, you'll see more upsets. Um I just think it's it's better across the board um for everybody involved. And the oh the other thing is um if you look at TV ratings or streaming ratings, what's the number one sport? Football. How many minutes of actual game time is there in a football game?

[00:50:41]12 15. If you look at ratings for sports, the less amount of game time, actual playing game time, the higher the ratings, people needs their chips and guac.

[00:50:56]And the NBA, I mean, if you look at hockey, hockey has the hardest time on TV. What happened with baseball ratings when they shortened the game?

[00:51:04]Straight up.

[00:51:06]>> Maybe they could take away all those timeouts at the end of the game. Well, I don't even mind the timeouts in a 40 because we did take away some of the timeouts, but if the game's only 40 minutes and there's just more action that counts, I think that changes the game.

[00:51:19]>> I had an idea that you could have an AI make a new rule every game. So, one game the three-point is actually worth four points. One game if you dunk, you're rejected. One game, you're just playing three periods, not four quarters.

[00:51:32]>> You tell me how that league goes and then I'll call you.

[00:51:36]I'm going to I think you're on board here. Um, last one, easy one. Uh, how do you fix the NBA draft?

[00:51:44]>> Um, I think, you know, shortening the games is one thing. See, people think it's about tanking. It has not even Even the NBA is wrong in this. They think it's about tanking. It's not about tanking. It's how do you rebuild a team?

[00:51:55]And to rebuild a team when you're bad, your payroll is lower.

[00:52:01]So you have more flexibility in trading picks. I think you make the game shorter. I think you limit the number of picks even more so that a team can trade from like it's for seven years. You limit it to not trade for three years and you eliminate the you um reduce the number of t number of first round picks that a team can own at every any one point in time because once you can't accumulate any more picks, there's no reason to stay under the cap. And you know if you can't stay under the cap, you might as well try to get better. So, that's what I would do.

[00:52:31]>> Love it. Uh, all right. I'm gonna turn it over to Dwayne. And I just wanted to say thank you to all of you. You've been amazing.