Transcription
Questions are being raised about the rapid growth of the AI sector. Some see a strong demand in artificial intelligence and technology. Others are concerned that we are in a bubble that will eventually burst. Here to talk more about those concerns is Gary Marcus, an author and professor emeritus at New York University. Gary, thank you very much for joining us today.
>> Glad to be here.
>> We've heard the term frothy a lot for this. Uh is it coalescing into a bubble or is it a bubble already to you?
It is a bubble. The question is when will the bubble pop? So the valuations just don't make sense in terms of the income that these companies are making uh and the enormous amounts of infrastructure that they require. It's just not economically sensible. It's all driven by hype and a kind of belief in a kind of false god of artificial general intelligence that could do anything but is actually pretty far from what we know how to make right now.
>> How do how how inflated do those do those valuations feel to you? I mean it it's hard to know but you have companies like OpenAI is losing a few billion dollars every quarter and its latest valuation is half a trillion dollars and it's made commitments for a trillion over a trillion dollars like it just doesn't really make sense. But if this if this is the start of something new, I mean something new for society literally, would that not be taken into account though that the the the chance that we're we're kind of pushing it a little bit, but the reality is we'll get to the next level.
>> I think we will, but I don't know if we'll do so in the next several years. I think there are a lot of technical problems with the current AI, and I've I've been documenting those problems for 25 years. Um, so I first wrote about hallucinations in 2001 and we all everybody now knows that they're a serious problem for LLMs, but I was pointing it out in 2019. Um, people keep saying, "Oh, we'll just add more data. It'll solve these problems." And those are always false promises. It's a little bit like Elon Musk keeps promising that driverless cars will take us from New York to Los Angeles next year. Um, and it never materializes. You know, someday we really will have driverless cars that good and we will have artificial general intelligence that good. It's just that we're not particularly close to it. The things that we have now are these black boxes that nobody knows how to debug, nobody really knows how to interpret, and everybody's just kind of hoping for the best. If I add more data, it'll all work out. And that just hasn't happened.
>> And where do you think where do we think you feel we are in the evolution of AR AI relative? I think a lot of people think next year with those drive driverless cars, but realistically, when do you think we're going to see all these things that they're that they're hyping right now?
So with driverless cars, you know, you can already take them in San Francisco, but you can't take them in 95% of the world's cities, and it's probably another 10, 15, 20 years before they become universal and ubiquitous like we were originally promised. Artificial general intelligence that you could trust to do anything you want is probably at least 10 years away. It might be 20 or 30 years away. They're basic ideas that I think we don't have yet. Like how do you have a system look at a complex world and build a model of that world that makes sense, that is stable, that it can make good predictions over? We we're still missing ideas. It would be like asking in the age of alchemy, when are you going to invent chemistry? When is biochemistry going to come along? You just don't really know because you're missing uh some basic things. So I I think the science is just relatively uh immature right now. People are pretending that it's much more mature than it is and we don't know exactly how long it's going to take.
>> And if this bursts, you think it's going to burst or we're in the bubble and if it burst, you think it's going to burst big and what kind of an impact could that have on the economy, do you feel?
>> Well, it's a question of how overleveraged we are. So, I think for sure some major investors are going to lose a bunch of money and the people who put money into those uh major investors um things like pension funds are going to get I think hit pretty hard. And then it's a question of like are banks going to get hit hard? Is that going to cause a liquidity crisis? There's a lot of information that's not publicly known um for kind of what the blast radius will be. I think it will be pretty serious. And if you look right now at the US economy, kind of the only thing that's happening is all of these drug um you know increasingly high valuations um in AI companies. And when that disappears, people are going to realize that things are not very stable. And then the kind of worst case scenario is taxpayers might be asked to bail everybody out. There's this this constant argument, oh, what if China gets these systems first, which I don't think is actually such a great argument because I think everybody's building basically the same thing. So, nobody's going to get a very big lead. But people keep using that as an excuse. And so, if all of this collapses, it's not hard to see someone like Trump saying, "Well, we should put in a billion dollars or trillion dollars, more like it, um, in order to rescue these companies." And that would come at the expense of the taxpayers. I hope that doesn't happen, but I can easily foresee it happening.
>> And what about the circular economy? Where does that play in with this?
>> Well, that's part of I won't use the word scam, but you know, it's part of the whole thing that gives me um the willies when you have companies buying services um from each other and basically it's all circular. Um it just doesn't really make sense. And we've seen that kind of thing uh happen before and it never ends well. You know, it's one of these things that drives retail investors up because they say, "Oh, there's all these bookings and stuff like that, but there's not a there there." And that's another reason to think it's all going to fall apart.
>> Okay. And but realistically, I mean, do you do you feel AI is here? It's here to stay and it will eventually reach its potential.
>> Yeah, I think AI is here to stay, but I would caution that what we're talking about now mostly is generative AI, these chatbots. And making a chatbot that's really reliable is extremely difficult, not something we actually know how to do at all right now. So the things that we're seeing now are still going to exist, but a lot of people are already starting to be disillusioned with them. They're using them in more limited ways than they imagined. Something will come along that's better. But like imagine that you were talking about helicopters in the age of Leonardo da Vinci and you said, "Well, are these helicopters going to, you know, are they here to stay?" And you know, I would say, well, you know, someday we'll have helicopters, but we're going to need advances in material sciences. Um, we're going to need advances in in engines and so forth. And then someday they will be ubiquitous when we make the right discoveries, which I don't think has happened yet. AI will really fundamentally change the world. Right now, you know, it has some utility for things like coding and brainstorming, but mostly it's hype right now. And it's going to be a while, you know, probably a decade at least before that hype is really cashed out into reality.
>> Okay, Gary, thank you very much for joining us.
>> Thanks for having me. Gary Marcus is an author and professor emeritus at New York.