Could an AI Bubble Pop and Take Your Job With It?
Some economists are worried about a troubling chain reaction: if the AI boom turns out to be a bubble that bursts, the resulting recession could actually speed up how fast companies replace workers with AI.
This idea comes from a Planet Money newsletter writer at NPR who says he has long doubted the biggest claims about AI changing the world. He is still not sure those claims are true. But two areas of economic research have made him take this "double whammy" idea seriously.
What this means for you
If you are new to all this, here is the plain version. A "bubble" means prices for something, like stocks, go way higher than what they are actually worth, based on excitement rather than real profits. When a bubble pops, prices crash and it can hurt the whole economy. This site exists to explain ideas like this without making you look up every other word, because nobody should need a dictionary to follow the news.
Here is the first piece of research. Back in 1978, economist Charles Kindleberger wrote a famous book about financial crashes throughout history. He found a pattern. A new technology shows up that really could change the world. Investors get excited, then overly excited. They borrow huge amounts of money to bet even bigger, building up debt. When reality catches up, prices fall but the debt remains, and that can spark a financial crisis.
This happened with railroads in the 1800s. That technology was real and powerful, but the bubble around it still crashed and led to a recession known as the Panic of 1873.
AI might be following a similar path today. A J.P. Morgan analysis found that just 42 AI-related companies made up about two-thirds of the S&P 500's stock gains since early 2024. At the same time, AI companies are borrowing hundreds of billions of dollars to build data centers and the power systems needed to run them.
This does not prove AI is a bubble. The dot-com crash in the early 2000s led to only a mild recession because there was less debt involved, unlike the 2008 housing crash. But tech analyst Benedict Evans has warned that competition between AI companies could push prices down so far that none of them make real profit, even while spending enormous amounts of money.
Now for the second piece of research, the one that connects a bust to job losses. You might think companies replace workers with technology slowly over time. But a study by economists Nir Jaimovich and Henry Siu found something different. They looked at jobs that were vulnerable to past waves of automation between 1990 and 2017. They found that 88% of the job losses in those fields happened during recessions or in the months right around them.
Put these two pieces together and you get an uncomfortable idea. If fierce competition keeps AI companies from making money, that same competition could make AI tools cheaper for other businesses to use. If that price war helps trigger a recession, businesses might use the downturn as the moment to cut jobs and bring in AI instead. Investors lose money on the bust, and workers lose jobs in the fallout.
To be clear, none of this is certain. We do not know if AI will ever make enough profit to justify all this spending. We do not know if a bust would cause a serious recession. And it is possible a downturn would leave companies with less money to spend on AI, slowing things down instead. AI could also end up helping workers do their jobs better rather than replacing them outright.
Still, there are ways to soften the blow if job losses do speed up. Unemployment insurance is one option already in place. Bill Gates has floated another idea: taxing AI and robots, which would make keeping human workers more affordable while raising money for safety net programs.
- Keep an eye on how much debt AI companies are taking on.
- Watch whether AI companies manage to turn a real profit, not just hype.
- Pay attention to safety net proposals, like taxing AI, that could protect workers.
- If your job involves repetitive tasks that software could do, start learning new skills now.
As a next step, check whether your own job tasks show up on any list of roles vulnerable to automation, and start building skills that are harder for AI to copy.
Source: NPR
