In recent months, prominent AI leaders and experts have issued a stark warning about the impending takeover of artificial intelligence in the job market: “YOU HAVE 18 MONTHS.” By the summer of 2027, they predict, AI’s rapid advancements will outpace human capabilities, potentially erasing up to half of all entry-level white-collar positions. Even the brightest minds may find themselves overshadowed by a “nation of geniuses in a datacenter.”
This forecast raises several concerns. First, it’s unsettling to consider becoming obsolete. Second, I struggle to take seriously predictions that resemble a secular apocalypse, where a god-like force erases entire fields of human endeavor with a mere gesture. Most importantly, this dire 18-month projection suggests that software will soon render human abilities useless. Yet, I see a more pressing issue: young people are already compromising their cognitive skills by relying on machines far before they lose their jobs to them.
I worry more about the dwindling capacity for deep thought among today’s individuals than the rise of intelligent machines tomorrow.
The End of Writing, the End of Reading
In March, New York magazine published a cover story that quickly went viral, not for its shock value, but because it articulated a widespread concern: students are using AI to cheat their way through education. With large language models enabling high school and college students to generate essays on any topic, teachers face a crisis in assessing students’ actual writing abilities. “College is just how well I can use ChatGPT at this point,” one student remarked. A professor echoed this sentiment, warning that many graduates will enter the workforce essentially illiterate.
The decline of writing is significant because writing is integral to thinking. This holds true for both professionals and students. In an editorial titled “Writing Is Thinking” in Nature, the authors argued that relying entirely on large language models robs scientists of the crucial process of comprehending their discoveries. Anyone who allows AI to handle their writing risks filling their screens with words while emptying their minds of genuine thought.
Reading is suffering even more. A pseudonymous college professor, writing as Hilarius Bookbinder, noted in a March Substack essay that “most of our students are functionally illiterate.” This is not an exaggeration. Literacy and numeracy scores are declining across the West for the first time in decades, prompting Financial Times reporter John Burn-Murdoch to question whether we have “passed peak brain power” just as we build machines to think for us.
In the U.S., the National Report Card from the NAEP recently revealed that average reading scores have hit a 32-year low, a troubling statistic given that the data only goes back three decades.
Americans are inundated with words daily—emails, texts, social media updates, and Netflix subtitles. However, these fragments require little sustained focus. Many seem uninterested in engaging with anything more substantial than a tweet. The percentage of Americans who read books for leisure has nearly halved since the 2000s. Matt Yglesias, in his essay “American students are getting dumber,” cited education writer Chad Aldeman, who pointed out that declines in 12th-grade scores are most pronounced among the lowest-performing students.
Even the nation’s brightest teens are largely avoiding longer texts. Last year, The Atlantic reported that students entering elite colleges often haven’t read a full book. Georgetown’s English department chair, Daniel Shore, noted that his students struggle to focus on even a sonnet. Nat Malkus, an education researcher at the American Enterprise Institute, suggested that high schools have fragmented books to prepare students for standardized reading-comprehension tests. In optimizing reading assessments, the U.S. education system may have inadvertently contributed to the slow decline of book reading.
The Serum of Literacy
The decline of writing and reading is critical because they form the foundation of deep thinking, as Cal Newport, a computer science professor and author, argues. The modern economy values symbolic logic and systems thinking, skills best honed through deep reading and writing.
AI, Newport says, is part of a long line of technological advances that threaten our ability to think critically. The rise of television coincided with a drop in newspaper readership and a decline in leisure reading. Subsequent innovations—like the Internet, social media, smartphones, and streaming services—have further diminished our focus. Research has shown that students using phones take fewer notes and retain less information, while “task-switching” between social media and homework correlates with lower GPAs.
“The combined effects of reading and writing are like a serum that enhances our capacity for deep thought,” Newport explains. “We need to keep taking that serum.”
Newport’s concerns echo those of scholar Walter Ong, who asserted that literacy is not just a skill; it fundamentally reshapes human thought and knowledge. Stories can be passed down orally, but complex concepts, like those in Newton’s Principia, require the written word. Ong noted that oral dialects often have limited vocabulary, while standard English boasts over a million words. If reading and writing “rewire” our cognitive processes, their decline threatens to unravel our intellectual capabilities just as sophisticated machines emerge.
Addressing the AI challenge isn’t as simple as avoiding the technology or banning it from campuses. AI is not a cognitive danger to be shunned. Research shows that models like ChatGPT can outperform many doctors in diagnosing rare illnesses. Dismissing such technology could lead to real-world failures. The line between when to use an LLM and when to refrain is often blurred.
This dilemma is clear in medical schools, where students are encouraged to use LLMs while remaining vigilant about advancing their skills alongside the technology. “I worry these tools will erode my ability to make an independent diagnosis,” said Benjamin Popokh, a medical student at the University of Texas Southwestern. “If all you do is input symptoms into an AI, are you still a doctor, or just better at prompting AI than your patients?”
During a recent rotation, Popokh’s class was tasked with diagnosing a case using AI tools like ChatGPT and OpenEvidence. The chatbots correctly identified a pulmonary embolism, but there was no control group, meaning no one worked through the case independently. Popokh found himself relying on AI after nearly every patient encounter, leading to unsettling moments where he realized he hadn’t thought about any patient independently that day.
In a viral essay titled “The Dawn of the Post-Literate Society and the End of Civilization,” James Marriott paints a dire picture of declining thought in mythical terms reminiscent of Edward Gibbon. He predicts that as writing and reading diminish in this machine age, our faculties for understanding will fade away, giving rise to a pre-literate culture with “the implosion of creativity” and “the death of democracy.” “Without the knowledge and critical thinking skills fostered by print,” Marriott warns, “many modern citizens may find themselves as helpless and credulous as medieval peasants.”
Perhaps he’s right. Yet it seems more likely that we won’t face an epic collapse of civilization. We won’t become violent or barbaric, nor will we inspire excitement in each other. There will be no grand narrative chronicling the decline of the mind, as the changes will be subtle. Leisure time will increase, home life will dominate our free time, and screen time will fill our home life, with AI content consuming much of that. As Orwell might have suggested, “Imagine a screen glowing on a human face, forever.” For most, this shift won’t even feel like a tragedy; we’ll lack the insight to mourn what’s lost.
Time Under Tension
… Or maybe not!
Culture is a reaction, and there’s still time to resist the pull of thinking machines and the quiet apocalypse of passive consumption. I hear the stirrings of this revolution often. The most common question I receive from parents about their children’s future is: What should my kid study in an age of AI? While I can’t prescribe a specific major, I emphasize the importance of valuing one crucial skill: deep thinking.
In fitness, there’s a concept called “time under tension.” Consider a squat, where you lower your hips from a standing position while holding a weight. With the same weight, someone can complete a squat in two seconds or stretch it to ten seconds. The longer duration is harder but yields greater muscle growth. More time equates to more tension; more pain leads to more gain.
Thinking similarly benefits from “time under tension.” The ability to patiently sit with loosely connected ideas allows for their synthesis into something new. It’s challenging to illustrate this principle by discussing others’ thought processes, so I’ll share my own. Two weeks ago, The Argument asked me to write an essay evaluating the claim that AI would take all our jobs in 18 months. My initial reaction was to dismiss the prediction as wildly exaggerated, perhaps leaving little to say aside from “nope.” However, as I reflected on the prompt, various ideas began to coalesce: a Financial Times essay, an Atlantic piece, an NAEP study I had saved, an interview with Cal Newport, and a recent gym thought about how time amplifies pain and gain. A framework began to take shape. I realized my article wouldn’t focus on technology displacing jobs, but rather on how humans are diminishing their own capabilities in the presence of new machines. We’re so preoccupied with how technology might surpass us that we overlook how it can also undermine our skills.
In disaster movies, we often see humanity unite to confront an impending threat. One might expect that the advent of all-knowing AI would inspire a collective awakening to our capacity for deep thought. Instead, I fear we’re preparing for this supposed super-intelligence by dulling our own faculties, retreating into a state of disinterest characterized by less reading, writing, and thinking. It’s as if some astrophysicists, convinced a comet is headed for New York City, decided to prepare by leveling Manhattan. That’s madness. We must not let discussions about the rise of “thinking machines” distract us from the true challenge of our time: the decline of thinking individuals.
