The Rise and Potential Bubble of AI Technology

The success of technology in the 90s led to a cultural movement, a gold rush, and a market bubble, similar to what may happen with AI now.

### Jeremy Howard Predicts a New AI Bubble
* 🌐 Jeremy Howard compares the current AI landscape to the late 1990s internet boom, where the success of technology led to a gold rush mentality, causing a **bubble that eventually burst**.
* 💸 The rising asset prices during the dot-com bubble created a secondary phenomenon of **increased market participation**.
* 📉 The dot-com bubble burst in 2000 when Intel's stock fell around **30% in one day**, leading to a **multi-year market decline**.
* 🧠 Howard predicts that the current AI boom will likely lead to a **similar bubble**, citing the early signs of **rapidly rising AI valuations**.
* 🛒 Unlike the internet's lack of immediate functional use cases at the time of the dot-com bubble, Howard points out that large language models like OpenAI's **GPT-3 have immediate utility**.
* Howard notes that while cryptocurrency has the potential to **change the scope of money**, there is currently no **imperative or immediate use case** for most corporations or individuals.
* 🏥 Howard observes a "**fear of missing out**" (FOMO) trend in various industries, such as medicine, where there is a growing interest in **leveraging AI technologies**.
* 📚 Howard mentions working on "**book Gen**," a tool that generates a **200-page book** based on GPT, showcasing the **practical applications of AI**.

really rapid, and you could kind of have blinked and missed it from a financial perspective, but you probably couldn't have blinked and missed it from a cultural perspective. There was, uh, I was a teenager at the time, and it was a really big cultural movement because there was, there were sort of two simultaneous movements, um, and one was, uh, causal from the other, but the general movement was the success of technology, and the culmination of that was, was the internet, and especially, um, as we, it sounds, America Online. America Online was an ISP that gave you this very weird wall. Garden, I would kind of call it a Walt alley, maybe more than a garden, of the internet, which was this weird, sort of limited, curated internet. And one of the things that AOL revolved around was this thing called keyword. So you type key, oh shoot, I got a Riz already, um, so anyway, long story short, uh, the, uh, the idea was, uh, is that I think we're going to go through something very similar. And the point I was trying to make was that the success of technology

in the 70s and '80s, sort of the software boom and things like that led people to the point where when we had a Communications internet, you know, boom in the in the 90s, that everyone sort of became a Believer. And then people could see it with their own eyes when AOL was out. So there was this willingness to sort of abandon all, you know, any and all kind of, um, I don't know, carefulness or, you know, prudence and, uh, start embracing risk for two reasons. One, that, you know, the success of Technology was so obvious. Bill Gates was sort of a, a household name. And that, you know, people didn't want to miss the second ride. There was a fomo of epic proportions. And the internet was so important and becoming quickly so important and so visible in everyone's lives that, you know, there was a gold rush. And there's no other way to sort of describe it but as a gold rush. So what's interesting is the rising asset prices created the secondary KnockOn phenomenon.

which is basically that everybody wanted to participate in the markets, and you saw an enormous amount of market participation. My family is from Albania. I have those Slavic genes like falic, and our family never invested anything in any market because, you know, they were from Albania, Eastern Europe. There's no stock market out there. You don't raise a family to think about stuff like that or 401k or retirement fund. And our family does all of that, but at the time, you know, it just didn't make any sense. So in fact, keeping money in the bank was already kind of like my parents were very suspicious. Coming from a communist country, that was the kind of thing that was necessarily a good idea. So they went ahead and saved a lot of money, and they did that by working really hard and working lots of jobs, and jobs that most people don't want to have sometimes. And you know, the point I'm making here is that I tried to convince

them to invest in the market. Me and everyone else in America, uh, saw the news every day. The internet's going to make you rich and that sort of, uh, FOMO led to a bubble. That bubble broke. We know the rest of the history. Starting in the end of 2000, Intel had a very, very bad, I think it was Q3, where they said they weren't going to meet Q3 expectations, not even close. In fact, it was so bad that Intel stock fell around 30% in one day. And it was, it was like a wake-up call to everyone that maybe the economy isn't going to go up forever, maybe there was, um, some pressure and some problem and some supply glut and stuff like that. And the whole kind of thing collapsed. And from 2000 to 2002, the market went down every year. It was a sea of red. It basically took until the Iraq War to sort of kick, kick the economy and the markets back into gear. And we know what happened since then. We had outside of

2008, we basically had a, you know, incredible, um, sort of, uh, incredible innovation and success, but most of that's been, um, fairly, how should I put it, um, uneventful. It's been incremental. This time is a bit, um, more like the 2000 time where it's a, it's a, it's sort of a discontinuous, you know, leap, and I think that we'll probably get a bubble. My point is that we're probably going to get a bubble very similar to the internet bubble, and I think we're starting to see the early signs of that, but I think, and I hope, that, you know, we're going to see something really, really gigantic. Uh, bubbles are, are, are good and bad, um, you know, most people look at them as, as sort of bad, but you know, there's, there's sort of this opportunity that can come with bubbles too. So anyway, um, you know, whether the AI, this new sort of AI bubble, will, will really happen or not, um, remains to be seen, but I do think and I feel like this

is the start of something similar to the 98-99 era where people started to think about how they had to be invested in all kinds of risky assets because, you know, they saw the wealth being created. And we're still very early there. It'll be very interesting to see the next Mark for open AI, whether that kisses $100 billion or more, or even if they ever raise money again. We're not really sure, right? But other phenomenon and other companies can raise money, and we saw yesterday or today Lang chain raise some money. I'm really happy for them as an early user of their tool and have written similar tools for my own use. And so, you know, I don't know what valuation they're getting, but it'll be really interesting to see if AI valuations start to go exponential or meteoric or parabolic or whatever hyperbole you want to use. And I'm excited because I do think that similar to the internet, we are seeing

a functional, immediately functional sort of core technology. The internet didn't have an immediately functional technology that was that impressive. There was the worldwide web, which was, you know, very enjoyable, but at the time, you know, other than GeoCities, Angel Fire, a couple of early kind of government sites, you know, there was not, not that much fun to be had. AOL, which is why AOL was kind of the killer app, you know, they introduced chat rooms, and you know, there was, there was a tremendous amount of fun to be had on AOL. Uh, but the internet really took time before it became really useful, and probably the most useful invention took a long time, which I would say cloud computing, um, you know, and it sort of took a while before SAS, Enterprise SAS, sort of came around, uh, and things like that, you know, Enterprise SAS didn't happen for at least sort of 10 years after the internet boom. So most of the really valuable companies that lasted were not like media companies like Yahoo or stuff like that. There were

there things that happened later like social media uh which happened well after kind of the bubble burst. Payments companies like PayPal which happened sort of during the tail end of the bursting and and so like you know the real Innovation the real money is sort of made after the bubble which is kind of funny but um it doesn't mean you can't be a bubble beneficiary. You look at Mark Cuban um uh Netscape. I was just looking the other day uh the history of the Netscape acquisition by AOL. You know there there there usually is some Foo on the corporate side too where the corporations kind of go a little uh a little crazy trying to buy their way into the next paradigm shift so I hope that you know we see this happen. So as I was saying a second ago unlike the internet which I know it sounds a little funny to say uh that the internet didn't have an immediate functional use case or functionality at the Inception but I think if you if you go back in time um

you wouldn't disagree with me. There were still a lot of people asking, What is this good for? and there were a lot of companies like Webvan, Cosmos, the Globe, and even one that my boss was running at the time called TheStreet.com. There were a lot of questions about, Do we really need this? And now, of course, we can't imagine our lives without the Internet. But at the time, you know, there wasn't YouTube, Google was still fairly new, there was no social media, so most of the things we use the Internet for today didn't exist at the start of the bubble. So, unlike that, we have OpenAI, which has this immediately functional product. I remember sitting in a jail and calling a friend of mine and asking him to go to a website called TalkToTransformer.com. And this was run on GPT-2, and it was the only website that really kind of ported GPT-2 API to the web. And I asked my friend to, he said

what do I type in this thing? And again, we're on this monitor jail call. It's 15 minutes in length at most, and I said, Why don't you ask it why did Carl Icahn and Martin Scy get into a fight? Now, I don't know Carl Icahn. I've never met him, and I doubt I'd have a fight with him. I'd probably fanboy out a little, but regardless, you know, he'd probably say something curt and nasty to me. But regardless, I'm a huge Carl fan. And my friend knows that I've never met him, don't know him. And so anyway, he typed it in, and I could hear his gasp. I could hear his jaw drop. He's like, Martin, this thing's writing the whole story. I said, I know. You could ask it anything, and it'll come up with this creative idea. He's like, This is insane. And I said, Yeah, it really is. Of course, I'm sitting in jail, so there's

not much I could do, but we, um, you know, we've seen GPT-3 and three and a half, Turbo and four, you know, all start to eclipse what we saw there. And my point is that we have immediate functionality in large language models that we didn't have really on the internet side, that that is useful right out of the box. It's useful right now, um, that should be enough to make a bubble as big as the internet, if not bigger. So I think that, you know, crypto also sort of suffered from this problem of cool technology, but where are we going with this and, you know, big, big potential use case, changing the scope of money and things like that, but no real great utility out of the box. You know, you still have to sort of suspend disbelief, um, that you have a store of value that that's not ephemeral. Um, you have to sort of ask yourself if this is really right for you, given, let's say you're a corporation or something like that. There's no immediate impetus to say if you're Johnson and

Johnson, or Proctor and Gamble, or Walmart, to say, Okay, let me change a bunch of our treasury to bitcoin. It doesn't really make sense right away, right? There's no, not just use case, but there's no imperative. There's no drive to sort of say, Let me change how I'm going to do things. And it very well, very well may be that crypto in 20 years, or 5 years, or 10 years, you know, who knows, becomes like a very standard thing for corporations and people to have, you know, 20 or 30% of their wealth. But as of now, there was no event that forced you to use it today. You know, you're sort of forced to use it, um, you're forced to use AI, or you get left behind. And I noticed in the New England Journal, um, and a couple of other medical publications that, and I'm curious if this is happening in other spaces, but I still pay attention to Medicine quite a bit. I noticed that there's, there's a

fomo going on where you know unprecedented in a sense that New England Journal has had all these uh articles about you know GPT and doctors. How do we use this thing? Is it good? Is it bad? Is it dangerous? You know, what exactly? And it's really uh an interesting kind of uh, you know, uh paradigm. You know, I've been spending a lot of time writing this thing called book Gen, which you can find in my GitHub, which is a little tool that writes a book. And it writes a 200-page book for you based on GPT.

Martin Shkreli: "AI Will Crash Like The Dotcom Bubble In Late 1990's"

Shkreli Planet

The Cultural Impact of America Online in the 90s

The success of America Online (AOL) in the 90s marked a significant cultural movement driven by the widespread adoption of the internet and technology.

AOL's curated internet and the concept of keywords led to a gold rush mentality and a surge in market participation.

The phenomenon created a willingness to embrace risk and a fear of missing out, contrasting with traditional views on saving and investment.

The 2000 Stock Market Crash and the Likelihood of a New AI Bubble

The 2000 stock market crash was triggered by Intel's Q3 performance, leading to a multi-year economic downturn until the Iraq War.

The speaker predicts a new AI bubble similar to the 1998-99 era, driven by the early signs of risky asset investment and the exponential rise in AI valuations, but with potential for both good and bad outcomes.

The speaker expresses excitement about the possibility of a significant AI bubble and its potential impact.

The Evolution of Functional Technology from the Internet to Cloud Computing

The internet initially lacked impressive and immediately functional technology, with the worldwide web and early websites like GeoCities and Angel Fire offering limited enjoyment.

AOL became the "killer app" with its introduction of chat rooms, providing a significant amount of fun for users.

Unlike the internet's slow start, cloud computing and enterprise SAS took time to develop but eventually became highly valuable, leading to the emergence of companies like PayPal and social media platforms.

Discussing the Power of Large Language Models like GPT-3

A conversation about using GPT-3 to ask about a fight between Carl Icahn and Martin Scorsese.

The speaker is impressed by GPT-3's ability to generate creative ideas and sees potential for large language models to create a huge impact.

The discussion also compares the immediate functionality of these models to the lack of utility in cryptocurrency, highlighting their potential for a wide range of use cases.

The Current Lack of Imperative for Corporate Adoption of Cryptocurrency

There is currently no strong reason or driving force for corporations to switch to using cryptocurrency for their treasuries.

The focus and pressure for adoption is currently on AI technology, as seen in the medical field.

The use of GPT and AI in medicine is creating a fear of missing out, leading to a surge in articles and discussions on its implementation.