In the 1600s, the Dutch had the beginnings of the first modern economy. They had banking, stock exchanges, and, eventually, even a futures market. Then it all came tumbling down in a frenzy over tulip bulbs.
You see, tulips were a new thing back then, an import from Turkey. Everyone, but especially rich folks, loved having gardens filled with these new, exotic blooms. The most prized flowers had striped, feathery petals, but these distinctive blossoms resulted from an aphid infection. That made them difficult to plant and grow. As a consequence, the law of supply-and-demand took over, and the price of bulbs went up.
The thing is, the price kept going up. You see, tulips were something people really liked and wanted. People started buying tulip bulbs right away, before the price went even higher. Another kind of law came into play, FOMO, or the Fear of Missing Out.
As the supply and demand and FOMO interacted, price increases accelerated. People into investing—remember, banking was kind of a new thing back then—came to see tulip bulbs as good investment. They started to buy the bulbs not to plant them and enjoy the flowers but as a way to make money. Soon, there was even a futures market, where they speculated about the future price of tulip bulbs. In essence, they were buying bulbs before they were available, betting that the price would be even higher in the future.
Then, suddenly, one day in February of 1637, traders didn’t show up when the market for tulip bulbs opened. The price for a single bulb had escalated to so high that it was comparable with that of a small estate house, and people were no longer willing to pay the price. Understand, everyone still liked and wanted tulips, just not at the now insanely high prices.
It all came tumbling down, and the market for tulip bulbs collapsed. All those people who thought the high prices would go on forever and invested in the bulbs lost fortunes overnight.
If this sounds similar to a Ponzi scheme, that’s because it is.
Ponzi Schemes and Bubbles
A Ponzi scheme is like a complicated chain letter. In a Ponzi scheme, the early investors receive as “returns” the money paid in by later investors. Of course, the late-comers think that their money is being invested and not just paid out as earnings to the earlier investors. As long as the number of new investors grows, the fake “returns” can continue being paid out. Unlike a chain letter, though, in a Ponzi scheme, there’s one master-mind who is cheating everyone and relying on FOMO and greed to keep the new “investments” coming in. It works until everyone realizes it’s a house of cards, and collapses.
In a Bubble, there’s no master mind behind the chain—it’s all supply-and-demand and FOMO. The Tulip Bubble of 1637 was one of the first examples of a naturally-occuring Ponzi scheme, where FOMO and greed eventually led to collapse.
In fact, “bubbles” are common in market economies. Remember the dot-com bubble? The internet was new and everyone saw endless possibilities. People liked going online, where information was free and varied. They wanted more of it, and hundreds of businesses started up to meet that demand. “Online” was just like “tulip bulbs”. People like it, wanted more of it, and the market responded. Supply and demanded interacted with FOMO. It led to a bubble and an inevitable collapse.
There are many other historical examples of bubbles, and all of them led to an inevitable collapse. The stock market in 1929 was certainly another famous one, but there was the sub-prime housing bubble (2006), the Florida land bubble (1925), the California Gold rush (1848), or even the Mississippi Bubble (1725) which collapsed the French royal banking system. These are just a few of the more famous. They happen all the time. One of the purposes of banking and investing regulations is to mitigate bubbles, until, of course, politicians vitiate the regulations. But, that’s a story for another time.
What does this have to do with AI?
Well, the current mania for AI has many of the hallmarks of a bubble. It’s widely regarded as what’s behind the current stock market, for example, which seems shrug off traditional negative indicators like consumer confidence and inflation, to say nothing of tariffs and war news. There is continuing, massive, and ongoing investment in AI-related infrastructure, from data centers to the companies making the specialized computer chips that drive AI.
But this particular AI frenzy is different, too.
For one thing, people hate AI. I don’t use the term “hate” lightly. As nearly as I can tell, people hate it with a particular white-hot passion. I don’t know about you, I’ve yet to meet anyone who wants more of it, let alone anyone who likes it.
So, in that specific way, it’s not at all like tulips, or low-interest and accessible mortgages, or the internet in general, all of which were things people liked and wanted more of.
So, who does like AI and want more of it?
That’s easy. Billionaires, for one. For another, large corporations. Both groups see AI as a way to cut costs, eliminate jobs, get richer, and exercise control over us serfs, er, I mean the public who worships them. You don’t even have to ask them. At least one billionaire posts regular tweets on his personal social media platform gloating over how rich his own (crappy) version of AI is going to make him by eliminating jobs for all us losers.
Still, even though people hate it, AI could still be a bubble, This time, though, it’s being imposed on the public, forcing us to use something we do not like and want less of.
Examples of AI I hate
A perfect example of this is Google. In case you haven’t noticed, Google is no longer a search engine, it’s an AI engine. Gone are the days when a “search” gives you a list of websites to choose from. Now, Googling instead gives you what sounds like a person answering a question. What Google is actually doing is looking at all similar questions in its “training data,” and summarizing them, in what sounds like a conversation.
The thing is, it’s not checking those answers for accuracy. In fact, the answers are often wrong. I can attest to recent queries about how to fix things on Mr. Gene’s Android phone (somehow, he sometimes accidentally does something unusual, and it’s not obvious to how to fix it—yet another story for another day). Anyway, Google’s answers almost always apply only to Apple phones even when I explicitly say “android” in the question. When I point this out, it “apologies,” says I’m right, and then usually gives yet another incorrect answer. After several tries, I sometimes get a usable answer, but more often I search YouTube. Its search engine is barely better, but at least it gives me videos to choose from—although what I actually want is a link to written instructions I can read. The point is that Google is no longer doing that, which it once did quite effectively.
By the way, I’ve started using the Brave search engine to go back a real web search, where I can pre-assess sources for reliability on own without AI “assistance.”
Another annoying example of AI being forced on me is the now useless “grammar checker” in Word, which is now more accurately a “grammar ignorer” or sometimes a “grammar messer-upper.”
Understand, I like technology in general. I’m an early-adaptor, the kind of guy who ought to like innovations such as AI. I’m retired, so my job’s not at risk. I’m not a Luddite who has a knee-jerk reaction against new things. The problem is that AI is not reliable and doesn’t work, at least not in the places where it’s being forced on me.
I also don’t want my utility prices to skyrocket because Amazon wants to put a data center down the road, so I don’t care for the infrastructure, either. At least in the Dot-Com bubble, the fiber build-out was taking place on existing pipeline rights-of-way and the tulip bubble would have given me something pretty to look at.
Is this an AI Bubble or Not?
It’s similar to past bubbles, but it’s different too, and not just that people hate it and want less of it.
Here’s another difference between the AI frenzy and past bubbles. Unlike tulip planters and dot-coms, there are basically only two big players right now in AI, at least in the US. There are lots of companies investing in the related tech and infrastructure, but only two real players in AI itself. As in the Dot-Com bubble or other earlier bubbles, some of these support companies will survive and even thrive. We got Levi Strauss and blue jeans because of the California Gold Rush, after all. But remember WorldCom? They bet heavily on the internet by stringing new fiber all over the US, most of which stayed dark for years. The company went bankrupt as a consequence, and almost took out the natural gas pipeline companies where they’d strung their cable.
Another characteristic of bubbles is that, in a bubble, investments exceed any reasonable expectation of return. Right now, AI implementations from the two leaders, Anthropic and OpenAI, are heavily subsidized, with estimates ranging from $3 to $25 for every dollar of computing consumed. The same is likely true for the in-house versions from other companies, or that they are crappy because they are done on the cheap. An unanswered—and, apparently, unasked–question is, will anyone, even billionaires, be willing to pay unsubsidized prices for AI products? It would be as though tulip bulbs really did cost hundreds of guilders. If that had turned out to be the case, no one would have ever purchased them.
Bubble or not, what’s it all Mean?
Not all AI implementations are the same as the two being pursued by Anthropic and OpenAI. The Chinese, for example, have models that use more limited training data and are less energy intensive, but get a high fraction of the results. If the Chinese version winds up prevailing, what does that say about the frenzy of investments in AI and related infrastructure in the US? Earlier bubbles gave us Levi Strauss, railroads, and fiber that we eventually used, but they also gave us the Great Depression.
Paul Krugman has written several blog posts on this topic, and I heartily recommend them to you. He’s a Nobel Laureate economist he doesn’t have the answer to where this is headed either, but he’s worth reading. Whatever it means, I’d trust him. I certainly would not trust an AI-generated answer!
Meantime, I still hate AI and want less of it. I do love all those tulips Mr. Gene planted in our front yard, though!
Be First to Comment