AI predicts patterns but lacks human understanding

Transcript

From the transcript The Diary Of A CEO publishes with the episode.

1. 31:18

   You do think we're in a bubble, don't you? I heard you talk about this. I think we're in a bubble, but not the same as the internet bubble back in 2000, 2001. I think what's different now is everything's a meme stock. Bitcoin is a meme stock, right? Pick Micron, you know, pick Google, Apple, they're meme stocks. And the reason I say that is by watching how stocks move with just a little bit of news.

2. 31:46

   So a company might miss earnings by a penny or only beat by a penny, and their stock will be down 18%. The intrinsic value of that company hasn't changed 18%. It's just that buyers and sellers move in and move out very quickly based off of anticipated information as opposed to the net present value of the future cash flows and all that. We're trading them like baseball cards. You know, there's no intrinsic value to this.

3. 32:13

   It's just perceived value. And what do you think somebody else will pay for it? And I think we're in a bubble for that very reason. What causes it to burst? I have no idea. Because we said, you know, this technology, AI is without precedents. You know, it is quite tempting for someone like me to put, you know, put a couple of dollars into the NVIDIA. Yeah, for sure. Because you think maybe this is the last technology.

4. 32:36

   Right, because even, you know, I don't think it'll be the last technology because it'll always, something will always happen. There'll always be somebody smarter than the AI, I believe. But look at Dell and Amazon. They were companies whose stocks went from way up high, got crushed when the internet bubble burst. Amazon was losing money hand over foot. I mean, people used to give them a hard time because they were losing so much money.

5. 33:01

   Now they're worth trillions of dollars. And if you were just stuck it out with them, so there will be winners. I just don't know which company. Is it Anthropic? Is it OpenAI? I don't know. You think there'll always be someone smarter than the AI? Yeah, for sure. What do you mean by that? I've never heard that. Take a model from Anthropic, AI model for Anthropic, or Google Gemini, or ChatGPT, OpenAI, right?

6. 33:23

   It gets released. It's immediately out of date. Immediately. The brilliance of this conversation is not incorporated in any model anywhere, right? AI is not smart. It just knows how to use statistics, basically, to predict the next word. It's predictive. It's not deterministic. Meaning if you asked... four different AIs, the same question, you're going to get four different answers. If you modify a question to the same AI, just marginally that shouldn't change the meeting, it'll give you a different answer every time.

7. 33:58

   That doesn't work in the real world. AI doesn't know the consequences of its actions. AI doesn't have inherent ethics. AI, you know, can't take video and consume it or absorb it, at least not yet. If I'm standing at a four-way intersection and there's cars going back and forth and I have video in my phone and, you know, the best models ever and I have a seeing eye dog right here. Mm-hmm. I'm letting the CNI dog take me across the street before I let the AI take me across the street.

8. 34:26

   It's always going to be behind. There's always going to be a latency. And so many things can change in a blink of the eye in this world that there's always going to be a human who says, uh, You know what? I can do this better because while all these other companies are competing with their models to try to, you know, hit these benchmarks and outperform in these benchmarks or do recursive self-improvement so it just keeps on training itself more and more, it doesn't have access.

9. 34:53

   A two-year-old sitting on a high chair with a sippy cup knows that if it pushes the sippy cup off the high chair... Mom's going to come running and it knows it's going to crack up, right? AI is not going to figure that out ever. Do you know, if you think of it as like an exponential technology, this is what I always try and... Because it's tempting to think of it as a sort of a static state. I'm thinking of like a graph here, like where it kind of goes like this.

10. 35:20

    But these technologies, from even what I've observed with every new release, seem to be almost like this. Yeah. So it's quite hard. Like, even think about full self-driving. Yeah. Like, my Tesla drives itself. And apparently, the crash rate is significantly lower than if a human was driving. Of course. And if I imagine that as an exponential, at some point, this shit's going to be pretty... Except.

11. 35:42

    Yeah. Except. They're just like you see the protests about data centers. Yeah. When we talk about why men rebel, a book I haven't thought about since high school, right? People are going to rebel. Yeah. People are going to create their own communities where there is, if it's that dominant, there'll be people say, I don't want that in my life. There'll be communes and kibbutzes where there's no technology.

12. 36:07

    There's something called adversarial actions for AI where like with full self-driving, there's certain things that when it sees it, it reacts in the wrong way. And once you know how to do that and you can cause AI to do things that it doesn't want to do or you don't expect it to do, then you're going to get people who do that all the time. And you're also seeing it like when I go and talk to hospitals and researchers for cost plus drugs, right?

13. 36:38

    What I tell them is you're crazy if you publish everything or if you try to patent everything because it's immediately available to every model and you lose control of your intellectual property. And you're starting to see more and more and more people instead of patenting it, it's a trade secret. Instead of publishing it, I'll put it in my model or I'll sell it to you. And so you get this bifurcation with each model where Gemini is not trained on the same data as ChatGPT, as Claude, as Grok, etc.

14. 37:08

    And if they're all giving you different answers— kind of defeats the purpose in a lot of respects. And if they can't deal with live video or they always have latency. And look, when I say this, I'm not saying AI is useless. No, exact opposite. AI is incredible. But like having four of the smartest people next to you, They're not all going to give you the same answer. And you have to have the skill set to be able to ask the right question and ask the right question of multiple sources and be able to take the output from each of them, the responses from each of them, and come to a conclusion that you think is in your best interest or whatever interest that is.