A Time for Conversation—and a Comment Calling Me a “Gangnam Leftist”
On September 3, 2026, Busan was still hot even though September had begun. I opened the broadcast as a time for conversation. Viewers could ask about Buddhism and scriptures, raise questions about another person’s video or book, talk about personal experience, or bring up social problems. When no question appeared immediately, I began with a recent comment that called me a “Gangnam leftist.”
I corrected it with a joke. I am not a Gangnam leftist but a Busan leftist, and I am not a wealthy Gangnam resident but a leftist from a poor background. I will not deny the label. I am progressive, a socialist, and closer to a democratic socialist. Of course, reality cannot always be divided neatly into right and left or conservative and progressive. Still, my position is clear: in a world where the rich grow richer, the poor grow poorer, and inequality widens, we must reduce that gap.
This problem becomes even more serious in the age of artificial intelligence. Changes in the labor market and unemployment have already begun, even if governments and institutions have not properly measured or announced them. Within a few years, both our lives and the world inherited by our children will change rapidly. If we discuss technology without asking whose productivity it increases, whose jobs it eliminates, and who receives the gains, we fail to see human life. In that sense, I do not hide my standpoint. I am not a Gangnam leftist but a poor leftist from Busan.
Microduck: Physical AI for $399
To warm up, I introduced Microduck, a small robot shaped like a duck. It costs $399, which I estimated at about 560,000 won using an exchange rate around 1,400 won to the dollar. It is not a massive industrial machine or a humanoid with two arms and five-fingered hands. It walks on two short legs, picks up objects with its beak, kicks a ball, and can move as if wearing roller skates.
What matters is that it is not merely a toy shipped with a fixed set of tricks. Developers can train movements in a virtual environment and apply the results to the physical robot. It is a small developer platform for experimenting with physical AI and reinforcement learning, while also being something amusing to keep nearby. Although its body and functions are simple, its relatively low price attracted a flood of orders.
Many reactions compared it to R2-D2 and the robotic companions of Star Wars. One commenter said that the dream of having a Star Wars-style sidekick was becoming real. Another liked the way it looked like a product imagined in the 1990s or 2000s. Someone else compared it to Pixar’s animated lamp. A tiny robot was pulling cinematic imagination into ordinary life.
I suggested that real AI robots may eventually be more capable than the robots in Star Wars. In the films, a human pilots the spacecraft while R2-D2 assists. Once AI becomes sufficiently capable, however, it will probably pilot better than a human. The same applies to driving. Robots walking beside us and helping us are no longer merely a distant fantasy.
Yet we should not stop at admiring a cute robot. If robots perform most of the work now done by people, large-scale unemployment will arrive with them. We must look at convenience and productivity together with possible labor-market collapse. The real choice is not simply whether to welcome or reject technology, but whether we will question how it transforms livelihoods and distributes wealth.
Homework Improves While Exam Scores Decline
The second story concerned the effects of generative AI on student learning. Its headline emphasized faster homework and lower exam performance, but the more important sentence appeared below it: the effect on learning depends less on the technology itself than on how students use it.
Some parents may prohibit their children from using AI because they fear it will make them intellectually lazy. I understand that concern, but I called a blanket ban a serious mistake that prevents children from adapting to a new world. It resembles forbidding computers after computers have become normal or demanding that every calculation be performed by hand when calculators are readily available. Workplaces already connect competence with the ability to use AI effectively.
This does not mean that students should hand an entire assignment to AI. When a teacher gives homework, a model can produce a result in seconds or minutes. That forces us to ask what homework now means. The same question reaches examinations such as Korea’s College Scholastic Ability Test, which rewards reading passages, choosing answers, calculating, and memorizing. If AI can already perform these tasks, why should humans compete to do the same things quickly?
When people must multiply large numbers, they do not normally write every digit on paper and calculate by hand. They use a phone or computer. Students should understand the principles of multiplication and division, why these operations are necessary, and how they are used. They no longer need to become faster than a machine. I argued that arithmetic education centered on rapidly performing addition, subtraction, multiplication, and division should disappear. Korea should long ago have moved beyond ranking people by scores produced through repetitive problem solving.
Tracking 26,811 Students for Thirty Months
The study described in the article followed 26,811 Chinese students between the ages of twelve and eighteen for thirty months. David Strömberg of Stockholm University worked with Victor Lei and Yanhui Wu of the University of Hong Kong. The researchers combined data from homework in nine subjects, monthly tests, and important entrance examinations.
After generative AI was introduced, homework scores rose by 18 percent and average time per assignment fell by about 30 percent, from sixty-four minutes to forty-five. AI compresses time. If a task that once required one hundred days can be completed in one, a result from the future has effectively been pulled into the present. I have often compared that effect to a time machine.
However, when the same students took monthly examinations without AI, their scores fell by 20 percent within six months, and their performance on important entrance exams also declined. Looking only at these numbers, one might quickly conclude that AI lowers children’s exam scores and should therefore be banned. I warned against precisely that conclusion.
If a student asks a clever classmate to do the homework, the assignment may earn a higher mark while the student learns nothing. Giving the entire task to AI is no different. The quality of an output must not be confused with the ability of the learner. The “metacognitive laziness” described in the article is a condition in which the student gains the product without doing the productive cognitive work. The decisive issue is not whether AI was used, but how it was used.
Students often used AI to obtain information and explanations of terms, but 31 percent used it to receive a complete solution to an assignment. About 29 percent used interactive or adaptive material as a learning tool, and roughly 20 percent used AI for personalized study plans or progress tracking. Many students employed it to escape unwanted homework rather than to pursue a self-directed question. Treating this outcome as proof that AI itself is harmful would be a statistical trap.
Change Learning and Examinations Instead of Banning AI
Learning with generative AI can give a person access to more varied and deeper information, faster understanding, and a greater volume of knowledge in the same amount of time. I imagined placing one thousand university professors from different fields in front of a single elementary-school student. Suppose the child becomes fascinated by grasshoppers or stag beetles and asks what they eat, how long they live, and how they reproduce. In real life, one thousand professors cannot remain beside that child and answer every question. AI can respond patiently and even provide images.
The same is true of Roman history, world history, science, and philosophy. We have entered a period in which a curious child can ask without end. School learning, by contrast, often forces the child to follow subjects and boundaries selected by somebody else. To score well, understanding is not enough; the student must study the examiner’s intentions, favored question types, and recurring topics. I do not call that genuine learning.
A person may study a field deeply and still receive a low score because the examiner selected different details. A poem can generate many reasonable interpretations, yet a standardized test insists on one answer. Students repeat questions with predetermined answers, and the score determines university access, jobs, and social rank. In an AI age, this method can no longer demonstrate a child’s real ability.
University Professors May Face What Human Go Players Faced
In the traditional Go world, famous players took disciples and transmitted both skill and style. Today AI plays better than the strongest humans. Human instruction does not disappear completely, but a student seeking the best moves cannot ignore artificial intelligence.
University professors may encounter the same situation. At present a professor’s knowledge and explanation may still surpass AI in some areas. No one can guarantee whether that advantage will last until the end of this year, next year, or several years from now. If everyone can use an AI capable of answering more broadly and accurately than one hundred professors at Seoul National University combined, people will ask why they must attend a university and listen to lectures in the same way as before.
I am not declaring that universities will vanish completely. I am saying that universities, professors, entrance exams, and evaluation cannot remain unchanged. Yet educational authorities often appear more interested in approving AI textbooks, devices, and products than in asking the fundamental question. Buying profitable textbooks and tablets is not educational reform. We must first change what students learn and what institutions measure.
Why Refuse a Child an Encyclopedia and a Thousand Professors?
When I was young, I saw a multivolume encyclopedia at a friend’s home. My family could not afford one, so I envied it and browsed through it whenever I visited. Parents once wanted to place an encyclopedia beside their children; today they should place AI there as well. Forbidding AI now is much like forbidding a child to use Google search after search engines have become part of the world.
For a few tens of thousands of won a month, or even free of charge, a child can access knowledge from countless fields. No parent could pay to station one thousand professors at home. The proper task is not to confiscate AI but to help the child follow curiosity, question results, and understand them in the child’s own words. Parents and teachers must demonstrate how that is done.
Questioning the Phrase “Fake Reading”
The third article, from The Hankyoreh, was titled “AI’s New Threat, ‘Fake Reading’—An Era That Requires Questioning Literacy.” I began by asking what “fake reading” could mean. If a person did not read the book, then the person did not read it; why call it fake reading? A user can ask AI to summarize a book in one thousand characters or explain the characters in the Odyssey. After reading 1984, the user can ask for an analysis of the emotional movement in Winston’s dialogue. In that case AI may deepen the reading. These different acts should not all be placed under one pejorative label.
Generative AI can process immense documents quickly and reshape them in the form a user requests. Feed it a book and it can discuss the contents without becoming tired of questions. Some tasks once possible only for a person who had read the work can now be performed in seconds, so the motivation to devote long hours to a book may weaken. That concern is understandable. It does not follow that human reading has ended.
Why I Could Not Read Ten Pages of Moby-Dick
As a high-school student, I tried to read Herman Melville’s Moby-Dick in a world-literature collection at home. I could not get through ten pages. The structure of a whaling ship, old nautical terms, and unfamiliar details of life in that period appeared in every paragraph. I was not a sailor and had no reason to know the name of every part of an old whaling vessel, yet those terms stood between me and the story.
Today I could ask AI about each unfamiliar expression and use images and explanations to continue reading much more easily. That may not destroy reading; it may lower a barrier that once blocked it. If a person wants to understand the story and its ideas, we should also ask whether forcing that reader to master every obsolete technical term first is really the best form of reading.
What Declining Reading Rates Fail to Measure
The article reported that the combined adult reading rate for printed books, e-books, and audiobooks fell from 67.4 percent in 2015 to 38.5 percent in 2025, and that only 28.8 percent of adults read at least one printed book. There may indeed be a decline in the amount of conventional text people read. Still, I asked whether the survey included webtoons and comics, even if web novels were counted. The way modern people read and acquire information has changed greatly, and measuring the present only with older categories of books misses part of that transformation.
Nicholas Carr warned that digital technologies weaken concentration and thought. His argument continues a long line of anxieties in which television was called an idiot box and smartphones, video, and social media were said to stop people from thinking. Generative AI now appears to go beyond distraction by performing parts of thought itself.
The Google Books Library project scanned more than twenty million volumes and made their contents searchable. Anthropic’s secret Project Panama reportedly purchased millions of secondhand books, cut them apart for scanning, used them for training, and discarded the physical copies. AI now processes quantities of text that no person could read in a lifetime.
Even if someone read one book every day for one hundred years, the total would be only about thirty thousand volumes—and no one can actually spend an entire life doing nothing else. How many people familiar with Francis Bacon’s four idols have read Novum Organum itself? How many people who know the story and symbolism of Moby-Dick have read the entire novel? Human beings have never directly read every original source. The claim that AI has suddenly ended an age in which everyone read everything romanticizes the past.
What matters is what a book says, what we can learn from it, and what kind of interest or pleasure it offers. People finish entertaining web novels, webtoons, and dramas even when told to stop. They also finish books that are absorbing and valuable. They abandon material that has no interest, no benefit, and only an excess of obscure terms. At that moment I joked that perhaps my own broadcast was unpopular because it was not entertaining. “Now I have truly realized it: I have to make it interesting.” It was self-mockery, but anyone announcing a crisis of reading should also ask why the content fails to hold the reader.
Humans Outsourced Thought Long Before AI
Books now warn of “outsourced thought,” “people who do not think,” “people who do not read,” the extinction of experience, and a crisis of reading. But failing to read and outsourcing thought are not identical. More importantly, humans were outsourcing thought long before artificial intelligence arrived.
People visited fortune tellers and asked monks about compatibility and auspicious wedding dates. They handed life decisions to someone reputed to possess special insight. Political judgment is assigned to politicians and commentators, science to scientists, and illness to physicians. There has never been an age in which every individual personally reasoned through everything in the world from beginning to end. Humanity has always delegated parts of thought to people and institutions.
The danger is not delegation itself but losing the ability to judge whether the received answer is true or false. If people live as AI instructs and lose even the standards by which to examine its guidance, AI becomes their master. More serious than not reading a particular book is never asking what I do not know, why I believe an answer, or what alternative judgments remain possible.
Apply the Four Questions for AI Answers to Buddhist Scriptures
In the article, literacy scholar Cho Byeong-young of Hanyang University argued that generative AI is changing the definition of literacy. In the past, literacy meant reading, understanding, and summarizing information accurately. In the digital era, comparing and analyzing multiple sources became important. The AI era demands another step.
- Can you doubt the answer produced by AI?
- Do you check the sources of that answer?
- Do you distinguish your own thought from the AI’s thought?
- Do you ask whether the information is desirable for the community?
I replaced “AI answer” with “Buddhist scripture.”
- Can you doubt what is written in a Buddhist scripture?
- Do you check whether the statement truly came from the Buddha?
- Do you distinguish the scriptural author’s thought from your own?
- Do you ask whether the teaching is desirable for people and communities today?
The moment these substitutions are made, they become religious criticism. In my view, most contemporary Buddhism does not ask them. People insist that everything written in a scripture is the Buddha’s word without investigating who wrote it, in what period, and for what purpose. They fail to separate the writer’s and translator’s ideas from their own, and they do not ethically judge the consequences that an old teaching produces in the present. The discussion ends with “It is true because it is in scripture.”
It is inconsistent to warn about hallucinations and mistakes in AI while refusing to admit any possibility of error in scriptures. People demand sources from AI but not from religious texts; they ask whether AI’s answer harms a community while assuming that an ancient religious sentence is right regardless of time and circumstance. Critical literacy is not a technique applied only to convenient targets. It must first be directed toward the authority that I most want to believe.
Buddhist Wisdom Begins by Refusing to Outsource Thought
The article called this responsibility “epistemic accountability.” I saw it as closely related to the Buddhist attitude needed even more urgently in the age of AI: question, observe, and contemplate. Buddhist wisdom does not emerge from handing thought to another person and copying down an answer.
A monk, scripture, professor, or AI can think on our behalf to some extent. We must not transfer the responsibility to examine that answer against our life and reality. Understanding the direction and context of history matters more than memorizing dates. Understanding the meaning of law matters more than reciting statutes. Asking whether a book’s argument is right, what assumptions it contains, and what consequences it produces matters more than merely summarizing its plot.
Therefore, we should not deprive children of AI. We should teach them how to use it well—not as a machine that completes unwanted homework, but as a conversation partner for pursuing genuine curiosity. They should learn to challenge its outputs, compare answers, verify sources, and restate understanding in their own words. Instead of confiscating AI because examination scores decline, we should first ask what those examinations, already losing meaning in an AI age, are actually measuring.
I may present an answer today and say a few days later that it was mistaken. That is not failure. What matters is not possessing a final answer forever, but offering today’s judgment and then observing again, correcting again. We must continue discussing what to demand from AI, what to demand from society, and why those demands are justified.
Artificial intelligence reads more books, summarizes faster, and calculates more accurately than human beings. That does not eliminate the human role. It enlarges our responsibility to decide what to ask, which answers to doubt, and what to judge as beneficial for a community. We should direct the same questions toward AI answers, school examinations, and ancient Buddhist scriptures. That is the attitude of a human being who has not outsourced the responsibility to think, and it is what I call Buddhist wisdom.
This article was edited by AI solely from the supplied video captions and description. No external sources were used.