AI Could Fix Higher Education by Breaking It

College degrees may prove worthless in the age of artificial intelligence. That’s not a bad thing.

Artificial intelligence has taken over the university. Students have seized on AI to get through college, using it to write papers, take notes, and do homework—along with structuring their social lives and relationships.

The ability to coast through college like never before does not bode well for students’ cognitive abilities or executive function. But there is also a great opportunity here. By making higher education maximally pointless, accelerating its uselessness to the zero point, AI could do swiftly what education reformers have failed to do for decades: overhaul higher education’s dysfunctions and open up the chance for a different, better equilibrium.

Last year, a quarter of students said they use AI to complete assignments, and a fifth use it to write complete essays. In a survey from the ancient past of 2023—the figures can only have gone up—nearly half of college students said AI had either performed a complete assignment or done the majority of the work, with the students tweaking it as necessary. It’s debatable whether AI use constitutes cheating, but half of college students themselves say it does, and the majority of college leaders say cheating has increased in the age of AI.

Is all this helping students master the material faster than ever? If so, they should be learning more each semester and coursework should be getting more difficult—the opposite of what’s actually occurring. Similarly, if AI facilitated additional learning, we would expect the number of hours spent on college coursework to remain level; instead, it is dropping.

Colleges have taken some steps to adapt, such as bringing back blue books and in-person exams. But by and large, higher education’s response has been a mix of acquiescence, denial, and shallow enthusiasm—revised integrity codes that feebly concede the inevitability of AI reliance, piecemeal reforms that will always lag behind student behavior, and superficial calls to “embrace AI” that ignore its fundamental challenges to the institution.

It seems, then, that while AI could be a boon to learning, its real-world effect instead has been enabling students to know less than ever, put in less effort than ever, and think less capably than ever, all while still earning that coveted college degree. In the words of a Claude-dependent New York University student: “I’m trying to do the least work possible, because this is a class I’m not hella fucking with.”

It’s tempting to conclude that AI is a disaster for learning. It is certainly bad for the status quo: better to write even a mediocre paper solo and learn something than to have ChatGPT write a pristine paper for you and learn nothing.

Yet precisely by accelerating this alarming trend to the point of collapse, AI may prove to be the key to higher education reaching a better place.

Slop In, Slop Out

Every week brings a disturbing news story or statistic about declining competence in the classroom. From 2019 to 2024, the share of 12th graders who were proficient in reading fell two percentage points to an abysmal 35%; math proficiency saw a similar percentage-point drop to an even bleaker 22%.

The number of 13-year-olds who never read for pleasure more than doubles those who do so daily—the reverse of the ratio only 20 years ago. Things are no better in college. One out of every eight freshmen at UC San Diego (which admits less than 35% of its applicants) cannot do middle-school math. Students at top universities cannot read a complete book. These findings are as troubling as they are bewildering: how can college students not know how to read?

This dismal trend comes as the cost of an undergraduate education has risen far faster than inflation, and as college’s practical value has become less apparent.

Scanning a random middle-school exam from a century ago will embarrass most of us today, no matter how many more degrees we have hanging on our walls. Give this trend a few more decades, and the average American will be in school until middle age to earn a doctorate indicating no more than the ability to spell his name correctly and add sums without using his fingers.

Appreciating AI’s potential to jolt higher education out of a lousy but persistent status quo requires first making sense of why that status quo exists. There is a curious tension here: how can we reconcile abysmal education outcomes with the fact that a bachelor’s degree still delivers a considerable earnings premium, and that college attendance continues to equate with individual accomplishment in the minds of most Americans?

After all, nearly everyone who has attended college in the last, say, several decades knows it involves plenty of nonsense—useless classes, head-scratching graduation requirements, bureaucratic box-checking (to say nothing of concerns about political correctness). And yet people still line up to attend college and are apparently better off for it in their careers.

Every troubling statistic and anecdote militates against the obvious answer, namely that people go to college to learn something. If the customers genuinely desired knowledge, they wouldn’t tolerate being skewered year after year by rising costs, falling standards, and grade inflation that lowers the value of their coursework as an indication of their knowledge or ability. In what kind of market does the customer ask for the product to be made worse?

In short, higher education has pathetically low standards and provides very limited preparation for the actual jobs students will enter, yet simultaneously benefits graduates in their career. So what’s going on?

Higher education is effectively a complicated signaling mechanism for employers; but the signal is only partly about knowledge or skills actually acquired through higher ed. What degrees largely signal instead is a willingness to spend time and effort on a given task, take orders, jump as high as one is told, and otherwise display traits of conscientiousness and conformity.

The ideal employee is diligent and deferential, willing to work hard on an assignment and go along with expectations set from above. These abilities are less tangible than the strict ability to perform a particular task, but no less important to being a good worker. Anyone can tell an interviewer, “I’m a hard worker,” but someone who spent four years writing papers on obscure topics can wave a diploma to show that it’s really true. The fundamentals of a degree—or, thanks to credential inflation, degrees—matter less than the more nebulous positional good it confers.

In short, students must devote themselves to increasingly costly signaling simply to remain in place relative to their peers in the job market.

But even as Americans sour on the college-for-all paradigm—the majority would prefer a three-year apprenticeship over free college—breaking out of it remains exceedingly difficult. No one individually can escape by declaring that the king has no clothes and a degree is basically worthless. College maintains enough luster in our society that even those who notice the absurdity are unlikely to opt out.

This is a major reason that higher education looks largely as it did a few decades ago despite many attempts at reform. What might work instead is a shock to the system—one that jolts us out of this rigid equilibrium and establishes a common understanding that much of the education system is bunk. AI could provide just that shock.

While it’s widely understood that college degrees don’t necessarily correlate with intelligence, AI could similarly poison the other signal—that of diligence and obedience. The allure of the essay-writing bot, after all, is that you don’t have to know anything or put in much effort to use it. There will be nothing of merit left for diplomas to reveal.

Adding to the shock value is that AI’s warping of the college model is happening in public, for everyone to see. What makes the technology so potent here—and different from the crushed dreams of yesteryear’s promised MOOC reformation—is that every employer will know that the college signal is broken, that the king has no clothes, and that every other employer knows it, too.

From the employer’s perspective, the realization that a degree no longer corresponds to intelligence or conscientiousness could prompt a desperate search elsewhere for hirable talent. Exactly how that might look remains to be seen; just as the college-for-all mentality solidified only over many decades, so too will developing alternatives take time.

Perhaps employers will return to older models such as apprenticeships and work portfolios. Aided by last year’s executive order weakening disparate impact liability, which made IQ tests a fraught enterprise for employers, perhaps more straightforward credentialing methods will emerge such as skills-based certifications or trial periods.

For this to happen, it need not be the case that every student relies on AI to cruise through college; some will surely refuse the low road. Rather, the college signal would just need to be broken enough that one isn’t sure whether a degree reflects merit or not.

A Profound Opportunity

What matters here isn’t AI’s possible use cases for education or anything else, but rather how AI is actually fitting into college education: as a tool to hack a system built for another time. Indeed, this fact is itself an indictment of our educational equilibrium. In a system that rewards performative hoop-jumping rather than substance, letting AI do the jumping for you is the natural move. It is precisely because colleges today offer education slop that they are so vulnerable to AI slop.

The accelerationist scenario in which the system collapses, allowing a new one to emerge from the wreckage, might seem ghastly on its face. But a wider historical perspective is in order here. Today’s education system, culminating in a college degree, is not some eternal representation of how learning has always been and must always be. Rather, it is the manifestation of continual adjustments to the economic and technological capabilities and demands of the day.

As the economic and technological underpinnings of American society evolve, it is only natural that our education system will need to evolve as well. Every major new technology shifts what skills are desirable and, therefore, what an education comprises; what is unique about today’s diploma arms race may simply be just how few skills even the “educated” have. Never has so much been spent by so many to learn so little.

Dismantling the higher ed establishment will be painful, but less painful than continuing down our present path for another generation. The light at the other end of the tunnel is a profound opportunity to put an obsolete system to rest and build a new one in its place—one that empowers students to start living their lives before hitting their midlife crisis.