hen Harvard is at its best, it approaches the ideal of a meritocracy.

However meritocratic it actually is in practice, Harvard and other elite educational institutions are nonetheless driving pistons and supporting pillars of our particular notion of an American meritocracy.

Churning elites and filtering talent as they do, they are essential to both the establishment and legitimization of large portions of the meritocratic elite — an elite that Thomas Jefferson called the “natural aristocracy” grounded in “virtue and talents,” as opposed to birthright or divine favor.

I propose the term “meristocracy” (apparently in some vain attempt to coin a neologism) to describe our particular American system where aristocrats derive their legitimacy from the idea that they rose up through the same meritocratic society as you and me — whether or not they attained their station purely by merit.

Most meristocrats possess fancy (and expensive) degrees that confer, like some artifact of the secular occult, the divine meritocratic right to pursue and assume positions as leaders of industry, academia, and government.

The social scaffolding upon which their status rests, however, faces unprecedented strain. Evidence now suggests that vanguard generations of AI models may soon have cognitive capacities superior to humans in almost every domain — particularly in lucrative sectors of the knowledge economy that long resisted automation.

So what will happen if — or more likely when — highly advanced AI models render the meritocratic aristocracy obsolete in the near future? That is: what if there is a great leveling of the playing field, and the differences in ability that long sustained the meritocratic myth, that had long underpinned the membership of Harvardians in the meristocracy for which Harvard is the chief symbol, are nullified?

In January, CEO of Anthropic Dario Amodei — whose company develops and runs the Claude family of large language models — published a 20,000 word essay about the promise and perils of superintelligent AI called “The Adolescence of Technology.”

In “Adolescence,” Amodei likens future powerful AI models to a “country of geniuses in a datacenter” that humans cannot hope to outcompete. If the current growth rate in the cognitive abilities of AI models continues, Amodei writes, “it cannot possibly be more than a few years before AI is better than humans at essentially everything.”

Amodei’s Anthropic now writes almost all of its own code with Claude. Models are now capable mathematicians, biologists, and financiers; they can give more accurate diagnoses than doctors and provide comfort as well as therapists. Machines are better drivers and manufacturers, more empathetic and available tutors, and faster (if not always better) editors.

Those jobs firmly in the corporeal realm are likely not much safer from automation: new AI models may massively accelerate robotics development, potentially enabling researchers to overcome the challenge posed by Moravec’s paradox — the observation that tasks comparatively simpler for humans are more difficult to automate.

Even if we take Amodei’s predictions to be overly optimistic — perhaps it actually takes 20 years to reach such intelligence levels, or AI will only be better than humans at 97 percent of economically valuable tasks — the effect may still be a massive upset in the knowledge work economy.

Traditional economic orthodoxy holds that technological advancements often arouse fears of mass unemployment, yet they eventually create new, previously inconceivable jobs or simply allocated labor to new tasks within the same industry.

The advent of ATMs brought fears that bank tellers would go the way of the dodo, but in fact allowed banks to open more branch offices, with bank tellers shifting to handle higher-skilled marketing and customer service work.

There is a strong case to be made, however, that in this Brave New World of large language models, human labor won’t fare so well.

The way Amodei frames it, “AI is increasingly matching the general cognitive profile of humans, which means it will also be good at the new jobs that would ordinarily be created in response to the old ones being automated.” Our adaptive abilities, in other words, may soon also be automated away.

It isn’t clear that human intelligence is so ethereal that there will always remain a domain unreachable by means of scaling laws — very basically, the empirical observation that more data yields better models. There exists the very real possibility that our minds will become economically and intellectually obsolete, or at least the relative differences that once magnified meritocratic separation will become irrelevant.

Even current AI models are predicted — after stumbling, error-ridden processes of implementation by aged conglomerates — to have labor-displacing effects, particularly for entry-level white collar jobs. The knowledge labor that AI doesn’t entirely replace may be “leveled” such that the relative abilities of a worker are not so important.

With the aid of a superintelligent model, someone who scored a 1550 on the SAT might be little more capable at writing legal briefs or fashioning slidedecks than someone who scored a 1320, for instance. This sort of dynamic is exactly what might undermine the current meritocracy.

Amodei discusses the labor shock that would accompany such predicted advancements of AI in the fourth chapter of his essay, entitled “Player piano.” Amodei doesn’t mention Kurt Vonnegut by name, but one must imagine that he is referring to Vonnegut’s 1952 dystopian novel by the same name.

“Player Piano” describes a future where highly advanced machines, designed and maintained by an elite upperclass of managers and engineers, have replaced nearly all human labor. Society is prosperous and none want for food or shelter, yet the lives of most are cruelly devoid of meaning and purpose.

Most people cannot contribute anything of market value to the economy and are helplessly dependent on the generosity of the state. Those that can — the engineers and managers — exist apart from and above the rest of society, breeding resentment and discontent in those forced into dependency.

These obsolete multitudes experience with painful acuity a dehumanizing humiliation, a loss of fundamental dignity brought on by no longer being able to live by their own hard work and effort.

We have witnessed what this looks like in living memory (well, not my living memory) in the American Rust Belt after the China shock of the early 2000s suddenly evaporated hundreds of thousands of indispensable manufacturing jobs. It is nothing short of traumatic to suddenly lose the ability to engage in what we have been taught is most virtuous: to work hard and provide for those that depend upon us.

How would Harvard students handle an abrupt technological advancement that rendered their hard-earned skillsets useless — or at least significantly less valuable? One cannot imagine that their psychology would fare much better than the other cohorts of workers that have stared down the money-greased barrel of automation.

Vonnegut didn’t foresee that the managerial class was little safer from automation than factory workers, that the same forces that stole the craft of even the humble barber in “Player Piano” would render knowledge workers obsolete in the modern economy.

Yet one must imagine that he would have a similarly pessimistic view of how an abrupt obsolescence would affect lawyers and programmers today: a loss of purpose, cratering dignity, a cruel humiliation served out on the order of millions.

Many a Harvard graduate aspires to work in this knowledge economy: according to a Crimson poll, in the Class of 2025, 21% and 14% planned to immediately enter the finance and consulting industries, respectively; another 18% intended positions in technology.

AI automation, however, is bearing down on “the Bermuda Triangle” (consulting, corporate law, and finance) just as it nips the heels of medicine, marketing, and middle-managing, to say nothing of programming. Entry-level jobs across all of these sectors are particularly under threat of automation.

Much of the justification for the meritocratic system typified by Harvard is that, though it may generate immediate inequalities (in that some receive more privileges and opportunities than others), it benefits society at large that we elevate the most talented people to positions of influence.

Their merit, their scarce ability to contribute a particular economic and cognitive value, justifies and legitimizes their higher status in society.

If indeed there is a great leveling, after which the kinds of skills and criteria that have long legitimized membership in the meristocracy are no longer so scarce, how then should we decide who deserves to enter the elite? Will the status of Harvard students still make much sense?

If the knowledge work that universities have long prepared students to engage in suddenly evaporates, or requires no special intellect to complete, what is left? That I even ask this question must say something about the pre-professionalism that infects many modern universities like Harvard, the need to justify every action in terms of ultimate economic utility.

College is viewed less and less as part of one’s development as an intellect, as a thinker and member of society, than as just a necessary step towards securing a job.

Even the humanities are now defended with such economic justifications (not necessarily with invalid logic): reading difficult classic books enhances your ability to read and comprehend difficult ideas in the workplace; analyzing literature makes for a more empathetic negotiator that can better understand his executive counterparts.

The humanities, some harried defenders remind us, simply develop us into better thinkers, lending an adaptable set of skills well-suited to all sorts of profit-maximizing roles.

That such contrived economic justifications could become moot might not be such a bad thing. The true value of the humanities probably does not lie in its ability to hew from rough timber lacquered corporate lawyers and management consultants.

Many of the jobs Harvard grads now vie for just aren’t very pleasant jobs, involving long, monotonous hours of corporate drudgery. In a world where it no longer makes economic sense for almost anybody to study finance or law, might we be freed up to devote their attention towards more meaningful, creative pursuits?

The particular manifestation of artificial intelligence diverges from what 20th century science fiction authors predicted in that AI is actually quite good at art, traditionally the abode of meaningful pursuits.

Indeed, AI can probably write a better poem or novel than most people could. Yet the existences of Patrick “Pat” B. Metheny, George Saunders, and Eliud Kipchoge do not affect the extent to which I derive meaning from playing the guitar, writing stories, or going on runs.

There remains the threat that a “Player Piano” like shock to knowledge workers would catastrophically disfigure the psychology of today’s winners of the meristocracy.

In any case, we must maintain our humanity. In a world where AI systems can do so much, yet humans remain the ultimate decisionmakers (for the sake of argument, I exclude the possibility of a system takeover), there is clear value in us remaining intelligent, learned, and thoughtful people.

Yet more important will be our morals: might future Harvard admissions select more for morality than intellect?

One can hope for a distilling of the intellectual and academic spirits that ought to lie at the core of a university, for a filtering out of the sulfides that rot bourbon curiosity. That might be something closer to what the true essence of a university ought to be. Somewhere in the soul of a university must be the inkling that, to quote “Animal House,” the 1978 National Lampoon comedy film that satirized Dartmouth’s fraternity system, “Knowledge is Good.”

In the eventuality of this radically altered landscape, where most knowledge workers can no longer contribute anything of economic value, the university must adapt. The meritocracy as Harvard has run it cannot simply idly abide the destabilizing tremors that lie ahead.

—Magazine writer Andrew W. Shlomchik can be reached at [email protected]. His column, “The Humanist,” explores what it means to be human in an era of unprecedented technological turmoil.