Like a lot of other people, I’ve spent the last couple of weeks playing around with OpenAI’s chatbot, ChatGPT. And I’ve been totally blown away.

I’m not alone, of course. More than a million people started using ChatGPT in the first few days it was publicly available, and my Twitter timeline has been filled with people posting screenshots of their favorite conversations.

ChatGPT is remarkably versatile. It can explain obscure philosophical doctrines, debug computer code, write and revise recipes, and compose stories on any topic—and in any style—the user requests. It remembers context from one question to the next, so if its initial answer isn’t what the user is looking for, the user can ask ChatGPT, in plain English, to revise it.

ChatGPT has gotten so much attention in recent days that I’m going to assume you are familiar with the basics. If not, I encourage you to read Sam Hammond or Maxim Lott or Jon Stokes. In this post I want to speculate on how this technology might impact the American economy in the coming years and decades.

The Hype Cycle

As a reporter I’ve always tried to be early in identifying and understanding technologies that will have a big economic impact. One thing I’ve learned over the last 15 years is that it’s difficult!

Take self-driving cars as an example. I wrote my first article on the topic in 2008. At that time I was excited about the technology but thought it was still decades from commercialization.

But then the technology advanced faster than I expected, and by 2017 I—like a lot of people—though we were just a few years away from large-scale commercialization. So when I took a job at Ars Technica that year, I made self-driving cars a major part of my beat. Since then, however, progress has seemed slower. Today I’m still optimistic about the technology’s long-term prospects but I’m unsure if it will be commercially significant before the end of the decade.

This pattern is so common that the technology consulting firm Gartner developed a fun model of it called the Gartner Hype Cycle. Self-driving cars reached the Peak of Inflated Expectations around 2018, and we’re now somewhere around the Trough of Disillusionment. With luck, we’ll ascend the Slope of Enlightenment and reach the Plateau of Productivity—where self-driving cars are a normal, boring part of the economy—within