Artificial Intelligence: the next Metaverse?
A few days ago on Facebook I made a joke about “Microsoft circling the AI drain”, and someone emailed to ask what I meant about it.
It’s simple: I wonder if generative intelligence and OpenAI will be to Microsoft what the Metaverse was to Facebook.
If you’re not familiar with the Metaverse (and why should you be, seeing how it vanished into obscurity), it was Facebook’s big idea for virtual reality back in 2020-2021. That was right around peak COVID, and it seems that Facebook’s leadership had the idea that the distancing requirements would be more or less permanent, that COVID had caused permanent change in the ordering of society. So it was time to go all-in on virtual reality headsets and software, since that would be the next major computing paradigm, in much the same way that first the Blackberry and then the iPhone had driven the mobile computing revolution.
This was a disastrous failure. Facebook ended up losing about two-thirds of its value, and its stock price dropped below $100 for a while for the first time in years. The company has since more or less recovered thanks to political ad spending on its platform, but not without a lot of layoffs along the way. If Mark Zuckerberg hadn’t set up Facebook so that it was nearly impossible to get rid of him, he probably would have been fired and a new CEO installed.
The three big problems with the Metaverse were 1.) no one wanted to use it, 2.) it didn’t make any money, and 3.) it cost enormous amounts of money to create and develop. Having something that costs a ton of money and doesn’t bring in any money is a big problem for any business in both the short term and the long term.
Generative AI doesn’t have the first problem – a lot of people want to use it for a variety of things – but it does have the second two problems in enormous spades. It doesn’t make a lot of money, if any, and it is enormously expensive to run the cloud computing infrastructure to maintain it.
Like, no one has really answered the question – how do you actually make money using generative AI? The hype around it seems to follow the usual tech investment bubble:
1.) Exponential user growth.
2.) ???
3.) Profit!
But tech entrepreneurs have a bad habit of ignoring step number two, which is unfortunate because it turns out step two is quite important.
(Midjourney is allegedly profitable since it offers subscription plans, though I wonder how much of the “profit” comes from outside investment.)
You can’t copyright anything AI puts out, which means you can’t use it in any business model that relies upon intellectual property rights, like publishing. Because hallucinations and mistakes are so common, you can’t use it for anything where you’ll get in trouble if it’s wrong. There was a famous recent case where a lawyer used ChatGPT to write a court filing, and ChatGPT cited six nonexistent cases. The judge was not pleased, and if you know anything about the American legal system, you know that having a federal judge mad at you because of your gross incompetence is not a good place to be. ChatGPT can generate code, but if you’re going to use that code for anything important, you really need an actual human to review it so you don’t get sued when something inevitably goes horribly wrong. So if you’re thinking to save money on developers by using ChatGPT, you still need to pay actual humans to check the code.
It’s very difficult to make money from generative AI, and the other side of the coin is the cost. Generative AI requires enormous quantities of computing power, and all that requires big data centers and lots of electricity. All of that is extremely expensive.
(As an aside, in a darkly amusing way, it is possible to make a lot of money using AI, or at least machine learning. YouTube’s and Facebook’s recommendation algorithms run off machine learning, and they’re so good at it that they caused a significant amount of global civic and social disorder over the last yen years.)
So on the one hand you have a product that doesn’t make much or maybe any money, and on the other hand is costs titanic amounts of money to produce.
How is this going to affect Microsoft?
On the surface (sorry, bad pun), Microsoft looks like a corporate empire in robust health. It’s consistently one of the world’s most valuable companies, and the stock price keeps going up. But if you pay close attention to Microsoft and its products, you can see the cracks starting to form. The company keeps having layoffs across all its divisions. The price for Xbox Game Pass just went up for the first time in a while. The much-touted Windows Recall feature was a security disaster and had to be, well, recalled, and that messed up the launch of its much-vaunted Copilot PCs. Windows 11 seems to keep getting worse and laden with more ads and bundled crapware. Speaking of security, Microsoft has been in trouble with the US government because of all the data breaches on Azure services. And speaking of Azure, “Azure is down” is something you hear more and more from people who work at companies that rely on Microsoft cloud services.
What I suspect is happening is that Microsoft’s leadership is throwing piles of money at AI projects since it helps drive the stock price up, but AI isn’t actually making any money, and the strain is starting to show across the rest of the company. So everything that actually makes money like Windows and Azure and gaming is getting cut to the bone so they can keep throwing money at AI since that’s what is currently driving up the stock price. But AI isn’t making any money, and eventually the whole thing is going to go bust.
This is all speculative on my part. And I don’t think Microsoft will go out of business or collapse when generative AI tanks, since the company has a lot of revenue streams, even if it is ignoring those revenue streams in favor of AI. I do think generative AI will do significant harm to Microsoft the way the Metaverse did to Facebook, but I could be wrong.
Nevertheless, in the last few years, we’ve seen a lot of speculative bubbles around what turned out to be totally useless and even destructive technology – crypto, NFTs, the Metaverse, and so forth – and a lot of the fervor around generative AI seems to be following a suspiciously similar trajectory.
-JM
