The Harvard Student Who Fed ChatGPT an Entire Novel

The Day a Harvard Student Fed a Novel Into ChatGPT

A student at one of America’s most selective universities got stuck on a novel. It was Anthony Burgess’s A Clockwork Orange (1962), and the student mistook its prose for “Old English.” So they dropped the entire text into ChatGPT to “translate” it into simpler language before actually reading it. A Harvard student.

When I first came across this story, I figured it was just one person’s literacy problem. But the more I dug into the data, the clearer it became: this isn’t about individual ability at all — it’s a sign that the act of “reading” itself is fracturing along social lines. Reading isn’t disappearing. It’s polarizing into a minority who read and a majority who don’t. You can see this divide in the share of people who read any books at all, and in how unevenly reading volume is distributed among them.

📊 Fewer Than Half of Adults Read a Book

According to a National Endowment for the Arts (NEA) survey, fewer than half of American adults read even a single book in 2022. Only 38% read a novel or short story. A study analyzing 236,000 records of how Americans spend their time found that the share of people reading for leisure on any given day fell from 28% in 2004 to 16% in 2023. For comparison, 57% of Americans placed some kind of bet last year. Because these surveys use different years and methodologies, this comparison alone can’t tell us definitively whether reading or gambling is the more common pastime.

The books people do read are also getting simpler. Sentences in New York Times bestsellers are about a third shorter than they were 100 years ago. In 1958, America’s best-selling novel was Boris Pasternak’s “Doctor Zhivago” — a work that probes the human interior through long, intricate sentences. Last year’s top seller, by contrast, was a new installment in the young-adult series “The Hunger Games.” Bestsellers have shifted from long, complex prose to short, easy-to-read prose.

We need to distinguish between the ability to decode letters and the ability to deeply read a long text. UCLA cognitive neuroscientist Maryanne Wolf argues that what people are losing isn’t the ability to decode individual words, but the higher-order capacity to hold onto a long text, understand it deeply, and synthesize it. She calls this “post-literacy.”¹ We’re reading more characters than ever through emails, texts, and posts — but our ability to focus on a single long piece of writing is declining.

In the U.S. reading data cited by the original article, the top 20% of adult readers accounted for more than 80% of all books read last year. That figure shows just how wide the gap is between people who read and people who don’t.

Why Reading Time Has Declined

Reading was never a natural human ability. Our brains have no dedicated circuit built for decoding text — we borrow the regions originally wired for speech and object recognition. Writing first appeared in Mesopotamia 6,000 years ago, and mass literacy only became possible after Gutenberg invented the printing press in 1440. In the sweep of human history, the era in which “everyone reads” turns out to have been a surprisingly brief exception.

Yet that brief era rewired how humans think, full stop. Media theorist Walter Ong argued in his 1982 book that writing created the very conditions for logical, linear thought. Neuropsychologist Alexander Luria’s experiments are the classic illustration. In the 1930s, he asked illiterate farmers in Uzbekistan: “All bears in the far north are white. Novaya Zemlya is in the far north. What color are the bears there?” Farmers who had learned to read solved the syllogism without hesitation; illiterate farmers refused to answer, saying, “I’ve never been there, so I don’t know.” Learning to read wasn’t just about decoding words — it was about gaining the capacity for abstract, logical thought itself.

The trouble is that other media have since taken that reading time away, one after another. Television came first; the internet and smartphones delivered the finishing blow. Marshall McLuhan predicted as early as 1962 that the West was heading toward a “post-literate” age, and Neil Postman warned in his 1985 book Amusing Ourselves to Death that TV had turned public discourse into entertainment. Things have only gotten worse since. There have even been reports that Netflix instructed its writers to assume viewers aren’t really watching the screen and to keep re-explaining the plot.

Schools, far from resisting this drift, have accelerated it. In a 2025 survey, most American middle and high school English teachers reported assigning only 0 to 4 books per year. As exams shifted toward short passages, classroom instruction shifted too — from whole books to excerpt drills. Classrooms are now saturated with digital devices. Meanwhile, short-form video has come to dominate everything. Psychologist Jean Twenge’s analysis shows that eighth graders (roughly age 13-14, the age of Korea’s junior high second-year students) spend an average of four and a half hours a day on social media — much of it watching video at 2x speed.

The marks this has left on the brain are unmistakable. Average screen attention span fell from 2 minutes 30 seconds in 2004, to 75 seconds in 2012, to roughly 47 seconds recently. Pediatrician John Hutton scanned the brains of children aged 3 to 5 and found that when watching animated video, they used only half as much of the brain region responsible for imagination as when they were shown a still image accompanied by audio. When video shows a child everything, there’s less left for the child to imagine on their own. Research has also found that the “Flynn effect”² — the decades-long trend of rising IQ across generations — has stalled or reversed in several countries over the past 20 years.

Now AI has been layered on top of all this as a new variable. If earlier technologies undermined reading, generative AI is the first tool to threaten writing itself. Georgetown’s Cal Newport argues that writing is thinking. The process of turning thought into sentences exposes sloppy logic and forges new connections. But when AI does that work for you, the ease comes at the cost of the learning that would have happened alongside it. In Brazil, undergraduates who studied with AI scored significantly lower on exams than those who didn’t, and one UK study found frequent AI use negatively correlated with critical thinking ability.

The Library of Alexandria is popularly remembered as having burned down in a single great fire, but modern historians point to a quieter cause: neglect. While damp and insects slowly devoured the papyrus scrolls, the will to copy and preserve them had already died out. Classicist Roger Bagnall argues that the very fact the library was allowed to die by neglect is itself evidence that a dark age had already begun. Today’s risk may unfold the same way — not as a single dramatic event, but as a slow erosion nobody bothers to stop.

🔑 Three Ways the Reading Gap Is Reshaping Us

First, our mode of thinking is regressing. Ong called it “secondary orality”³ — the return of literate societies to the habits of oral culture. Oral cultures lean on repetition, epithets, emotion, and conflict to aid memory. They grow numb to contradiction, and who is speaking starts to matter more than what is factually true.

This connects to politics, too. The media theorist Joshua Meyrowitz argued in 1985 that electronic media had shifted the standard for choosing leaders from a “résumé” to a “date.” Someone who looks good on screen gains an edge over someone who writes and argues well. Algorithms reward simple, provocative, emotional content over complexity or nuance. The original article points to President Trump as an example of this structure. But I think it’s more accurate to read this as a problem of media structure than a problem of any one individual. As the article itself concedes, politics built on short-form video and algorithms cuts across left and right alike.

Second, reading itself has become a marker of class. As we saw earlier, 20% of adults accounted for over 80% of the books read last year. Carrying a book around now reads almost like a “declaration of identity.” It’s gotten to the point where online communities mock the practice as “performative reading.” Leah Price, a historian of reading at Rutgers, says reading is turning into “a niche hobby, like stamp collecting or growing orchids.”

Third, cultural power is shifting. Newspaper employment has fallen 75% over the past 20 years, and humanities degrees have plummeted. In their place, the center of culture has moved to streamers, podcasters, and influencers. Podcasters with massive audiences, like Joe Rogan, now wield outsized influence. In the words of the social psychologist Jonathan Haidt, knowledge used to flow vertically, from older generations down to younger ones — now it flows horizontally, from peer to peer. About 60% of Gen Z say they’d want to become an influencer if given the chance.

Of course, I want to give the other side of the story a fair hearing too. Every era, in fact, has voiced similar anxieties about new media. Back in 1900, The Atlantic lamented that “newspapers are crowding out books,” and Jefferson believed novels corrupted people. Even now, print book sales have risen compared with a decade ago, new bookstores are opening, and Substack and audiobooks are growing. After one Dallas school district banned phones, library checkouts rose nearly 25%. Still, one detail nags at me. Even Martin Puchner, the Harvard comparative literature professor who has long insisted that “doomsaying is always wrong,” admits that this time, a return to text looks unlikely.

Oswarld’s Lens

Honestly, I think the phrase “the end of reading” is only half right. The statistics don’t show reading disappearing — they show readers splitting into two camps: those who read and those who don’t. The 20-80 skew — where 20% do 80% of the reading — brings to mind a Pareto distribution⁴. That said, one ratio alone can’t confirm the exact shape of the distribution, or whether the gap is widening over time.

There’s a pattern I’ve seen again and again while building market-entry strategies. The moment a behavior shifts from “mandatory” to “optional,” the market almost always splits in two. A dedicated minority who keep doing it, and a majority who barely do it at all. It happened with exercise, it happened with cooking, and now reading is heading down the same path.

So the real question isn’t “will reading survive?” It’s “who stays in that persistent 20%, and what kind of power does that minority end up holding?” In the AI era, as the number of people who read deeply and think for themselves shrinks, the value of that ability only rises. As Newport put it, deep thinking becomes more important precisely as it becomes rarer. No matter how smoothly AI churns out text, the job of discerning what’s true and thinking things through for yourself ultimately stays with humans. That judgment is exactly the edge the persistent 20% will hold.

Let me add one honest caveat. I’m wary of treating all this as some “end of civilization” narrative — a century ago, people said newspapers were killing books, too. But the fact that even the skeptics are starting to say this time is different isn’t something I can easily wave away.

Closing

First, reading isn’t disappearing — it’s concentrating in fewer hands. 20% of adults accounted for over 80% of the books read last year. Second, that gap is reshaping not just habits but where the center of political judgment and culture itself is shifting. Third, I believe that the more AI advances, the more valuable it becomes to read deeply and think for yourself.

Just as some read the decline of the Library of Alexandria as a story of neglect rather than catastrophe, the effort to keep reading alive matters just as much today. The threat we face now may not be some dramatic event, but quiet indifference. Fortunately, unlike that library, our books aren’t disappearing. Whether we read them or not is entirely up to us.

If there’s a book you finished in the past 6 months, drop the title in the comments. If not, tell me why you couldn’t get into anything. In the next issue, I’ll pull together the common patterns in where readers say they stalled out.

---

📨 If you know someone who’s been struggling to pick up a book lately, send this their way.

---

References & Further Reading

Primary sources

• The Atlantic, “The End of Reading Is Here”, 2026. : This is the original article behind today’s newsletter. I found its arc — starting at the Library of Alexandria and ending with AI — genuinely striking.

Background

• Walter J. Ong, Orality and Literacy, 1982. : This is where the concept of “secondary orality” comes from, along with an account of how writing reshaped human consciousness.
• Neil Postman, Amusing Ourselves to Death, 1985. : This traces how television turned the public sphere into entertainment. Reading it now, it feels like it’s describing smartphones.
• Marshall McLuhan, The Gutenberg Galaxy, 1962. : This is the book that first coined the term “postliterate.”
• Joshua Meyrowitz, No Sense of Place, 1985. : This examines how electronic media reshaped the standards we use to judge political leadership.
• Plato, Phaedrus. : This contains the passage where Socrates worries that writing “will ruin memory.” It’s the same debate, just a 2,400-year-old version of it.

---

The author is Oswarld (Kwangseob Ahn). Current roles: Adjunct Professor at Sejong University, Strategy Consultant at INLEVEL9. Career, research, books, and recent work are kept current on the About page. Latest · July 2026: HEMA-2: A Consolidation-Aware Tri-Memory Architecture with Multi-Channel Scheduling for Lifelong Conversational AI.

📝 Glossary

Footnotes

1. Postliterate: this doesn’t mean being unable to read — it means the capacity to deeply understand and synthesize long-form text has weakened, even though basic reading ability remains intact. It’s not the opposite of illiteracy; it’s the stage that comes after literacy. ↩

2. Flynn effect: the phenomenon of average IQ steadily rising from one generation to the next. But over the past 20 years, studies in several countries have found this rise stalling or even reversing. ↩

3. Secondary orality: the tendency of societies that have passed through print culture to revert to the habits of oral culture — repetition, emotion, improvisation. Ong coined the term to describe the age of broadcast and electronic media. ↩

4. Pareto distribution (power law): a lopsided structure in which a small minority accounts for most of the total. “20% of people do 80% of the reading” is a textbook example. ↩