The rise of AI has only added to doubts of a degree’s value. But some economists say college still holds some implicit value, like teaching students things AI could never learn how to do.
“If AI makes these jobs easier, you will see more activities shifting towards places where labor is cheaper, whether that’s India or the Philippines,” Frey told Fortune. “I think that’s going to put a lot of pressure on people’s wages doing knowledge work.”
Despite his estimation, Frey said earning a college degree is still worthwhile, as it imparts three core skills in which humans hold a competitive edge over AI: complex social interactions, creativity, and navigating complex environments.
AI has made leaps in communication advancements during the past decade. Despite that, Frey said those improvements actually make human-to-human interaction more valuable.
“The value of social skills have gone up over the past decade, whereas the value of math skills has been trending downwards,” Frey said.
Sure, you can ask ChatGPT to read the Rolling Stones’ “Sympathy for the Devil” in the manner of William Shakespeare, or even train an algorithm in impressionist art and ask it to turn your wedding photos into Monet paintings. But human creativity extends beyond memorizing knowledge and regurgitating it in different formats. It takes the ability to think differently and push boundaries.
“If you had asked an LLM in 1900, ‘would humans ever be able to fly?’” Frey said, “it would have concluded that there’s no bird that weighs more than 30 pounds that’s able to get off the ground.”
Frey said AI doesn’t quite possess the resiliency to function like a human. It can provide—with the click of a button—a wide range of information, from a slew of complex legal cases to optimized travel itineraries. But it doesn’t do well in environments that are in constant flux, as is the real world.
“An undergraduate textbook will not have changed that much in recent decades,” Frey said. “AI thrives as a tutor in those relatively static environments.”
“In professions where you have more volatility where your job changes more day to day, [those jobs] are less likely to be exposed or automated,” he said.
A version of this story was published on Fortune.com on Jan 14, 2026.
