This Andrew Yang tweet was going viral:
Heard about a small college in the Midwest that has lost 50% of its enrollment in the last 10 years. Hundreds of colleges are going to close in the next decade and it’s going to be devastating for many communities.
— Andrew Yang🧢⬆️🇺🇸 (@AndrewYang) August 30, 2026
Whether it’s hopes or promises of AI making white-collar jobs obsolete (wrong) or college obsolete, people keep projecting their desires, when the actual data doesn’t support it. There is no evidence yet of college being devalued due to AI. I have long maintained that AI is not going to change this much, especially for top-20 schools. Colleges will adapt, if they haven’t already.
So far I have been right, in posts such as “AI will not hurt the signaling value of elite colleges,” and “In an era of overabundance, elite colleges matter more than ever.”
Credentialism sucks, but the alternatives do not scale as well. Employers, especially for high-paying jobs, are inundated with applications. Elite degrees still signify outlier ability, making them valued in the eyes of employers, and AI has not yet changed this. Companies get thousands of applications, and there is no possible way to interview everyone, hence the need for cheap and legally compliant filtering at scale. When interviewees use AI, companies adapt. There is a back-and-forth process, with neither side ever gaining the upper hand.
But to address the tweet, it’s not collapsing though. “Around 100 or more four-year institutions have closed in that time.” But there are 3,400 to 4,000 degree-granting institutions in the United States according to the National Center for Education Statistics. So, statistically speaking, we would expect some annual churn, just as restaurants or other businesses open and close. The longer trend is what matters.
To try to settle the matter, I consulted Claude: the supposed collapse of higher education appears unfounded, although male attendance has dropped, but that is a separate issue:
People may be misreading a “decline of male attendance” to mean a decline overall, but this is wrong. Of course, there could be an enrollment cliff due to AI, but the evidence doesn’t yet bear it out. Why not just wait for the actual data to confirm it, than keep making the same tired prediction, if it’s so self-evidently true?
@Tyler_A_Harper of The Atlantic, chiming in, also went viral, entertaining the possibility of a sudden collapse of higher ed:
People are under-pricing the possibility of a black-swan-esque event where significant chunks of higher ed collapse suddenly. At some point parents will decide they will not pay $100-400k combined for their kids to cheat their way through college. I expect a bang, not a whimper. https://t.co/qXl7VcbcPr
— Tyler Austin Harper (@Tyler_A_Harper) August 31, 2026
“At some point,” as in never. When do you just admit you were wrong, rather than digging into the same failed predictions? The college wage premium is still significant (although it has fallen a bit since 2022) despite AI and student loan debt. That is why parents keep paying up.
As in the case of Tyler Harper, who is employed by The Atlantic, clearly getting a degree helped his career. Would he trade it away for a job in “the trades?” Likely not, but he thinks others will? Somehow parents will be like, “Yes, you should totally forgo the extra income and status that comes with a degree for menial work instead, because of AI…that is the ticket to success.” Like it or not, a degree is still the only game in town for being taken seriously by other important people.
This seems right. If you’re a parent with kids in the near-college age, it likely seems risky to not send your kids to college. But it ALSO seems risky to bet that $75-100k tuition will payoff. So I imagine a lot of people will hedge and go state school.https://t.co/EKGIzT782B
— Tyler Austin Harper (@Tyler_A_Harper) August 31, 2026
It’s closer to $35-40k, and the college wage premium makes it worthwhile nevertheless. Just run the numbers: the student loan payments are small relative to earnings, and fixed relative to inflation. He’s also moving the goalposts from “college/higher-ed in general will collapse” to now “state schools will survive.”
But I think it’s the opposite: elite schools will thrive, but ordinary/lower-ranking schools will need to adapt to AI. Because AI lowers the barriers to knowledge, low-ranking degrees are possibly more fungible or devalued. This is why the MIT open courseware program, despite lots of hype over the past 15 years, never caught on. As it turns out, an online certificate of completion is no substitute for an actual MIT degree.
Overall, much of this discourse–whether it’s failed predictions of AI being a fad (which 3 years later, it’s clearly not), a bubble (maybe, but so what?), to now colleges or white-collar jobs being obsoleted–comes off as “cope”. Seeing white-collar jobs vanish or degrees get devalued helps people feel better about not having a high-paying job or an elite degree.
Society doesn’t conform to your notions of fairness. AI likely isn’t some great arbiter that will rectify the inequities of modern life. All historical evidence shows technology only accentuates individual differences in talent and widens inequality. Expecting AI to be the exception isn’t supported by the evidence.