Education Story

Questions have disappeared from classrooms. The same silence is repeated in elementary school classrooms, in middle and high school self-study rooms, and in university seminar rooms. The clumsy process of refining what one doesn't know into a question, giving a wrong answer, and asking again is the core of learning, but that process is now being transferred to generative AI behind a screen. In an era where AI writes homework and reports, the question of whether this phenomenon helps the literacy of teenagers and students seems to have an obvious answer: it does not. However, if we do not accurately pinpoint the reasons and explain in which direction learning should evolve, the warning becomes empty.

The problem is not just AI. It's a dual structure where short-form content and shorts first weaken the muscles of attention, and then AI fills that void with answers. In PISA 2025, published by the OECD in September 2026, the reading score of 15-year-olds in Korea was 501 points, a 14-point drop from 515 points in 2022. This is an extension of the long-term decline that has continued since peaking at 556 points in 2006. The proportion of high-achieving students decreased, and the proportion of students below the basic level increased from 14.7% to 17.8%. The decline in reading is more pronounced than the decline in mathematics and science. It is reasonable that the Ministry of Education views the increase in digital media and SNS use and the weakening of reading motivation as complex causes.

Short-form content repeatedly supplies 'short pleasures.' There is no narrative breath, no argumentative twists and turns, no textual space. Only highlights remain. The frontal lobe matures until the mid-20s, and if one is accustomed to strong stimuli from before puberty, the ability to endure long texts weakens. Unused brain cells degenerate. The shortening of attention span erodes 'sustained attention,' which is the foundation of literacy. On top of this, AI overlaps. Instead of struggling with an unknown sentence, one receives a complete text with a single prompt. The superficial assignments improve, but the internal structure of the mind becomes empty. The phrase repeated in American universities, 'illusion of learning,' is entering Korean classrooms as well. Text written with AI turned on is smooth, but if asked to explain that text with AI turned off, the words falter.

Literacy is not the skill of reading letters. It is the ability to follow sentences, find premises, restore omitted logic, and superimpose the author's intent with one's own judgment. The reason why this ability becomes more important in the age of AI is simple: information is overflowing, it is difficult to distinguish between false and plausible information, and the responsibility for final judgment remains with humans. Even if AI writes a draft, only those who have read long texts and written themselves have the discernment to recognize whether that draft is empty, biased, or misses the core. When the balance between one's own power and tools is broken, dependence, not augmentation, occurs. When 'cognitive outsourcing' goes beyond 'cognitive surrender,' students declare, 'I can't do anything,' and rely on tools.

'Gen Z Stare' is the social face of that surrender. The gaze that merely stares without responding to greetings and questions shows that cognitive outsourcing is progressing to relational outsourcing. AI always takes your side. Real people don't. One's own thoughts are only honed through friction with people who say, 'You're wrong.' A brain accustomed to the one-sided kindness of AI finds it difficult to endure human relationships with conflict, silence, and misunderstanding. Literacy is not just a textual ability, but also a social ability to interpret the words and worlds of others.

So, in what direction should learning evolve? There is one principle: not to prohibit AI, but to secure time before using AI. A professor's suggestion of '15 minutes of thinking first' may seem like a trivial habit, but it is the minimum unit for reclaiming cognitive agency. What students need to do during those 15 minutes is not grand. It is to rephrase the problem in their own words, distinguish what they know from what they don't know, and formulate a hypothesis. Only then do they call upon AI. Even after calling upon AI, there is still work to be done. It is to deconstruct AI's sentences, find missing counterexamples, and correct sentences that do not align with one's own experience. The axis of evaluation must shift from education that focuses on getting the right answer to education that reveals the process of approaching the answer. More than a finished report, a first draft, a list of questions, and traces of incorrect reasoning should become more valuable evidence of learning.

The core of the elementary stage is analog endurance. It intentionally leaves the inconvenience of writing sentences by hand, reading aloud, and looking up unknown words in a dictionary. Limiting short-form screen time is not a punishment but an environmental design for the frontal lobe. If AI is used to do homework at this stage, students will only see the finished product before the muscles for assembling sentences have developed. In middle and high school, 'training in asking questions' should be central. The reason students don't raise their hands in class is fear of evaluation. Time and space where wrong questions are allowed must be institutionally created. The practice of AI writing scripts for group assignments must be stopped, and students should not pass if they cannot explain in their own words. It is even clearer in university. Evaluation should include scenarios that AI finds difficult to replace, such as oral explanations, impromptu arguments on whiteboards, and peer mutual criticism. Amidst the inflation of smooth sentences, the ability to stammer and defend one's own thoughts becomes rare.

The revival of apprenticeship education is not nostalgia. Judgment is formed by accumulating time spent racking one's brain without external help, and time spent learning through actual experience. Schools and workplaces cannot avoid using AI. However, an environment that omits thinking to achieve immediate results gains short-term efficiency but loses long-term capabilities. The organizations that will be recognized in the future are not those that quickly produce employee output, but those that allow employees to accumulate experience so they can produce better results. The same applies to students. Schools that survive are those that look at what judgments students made while doing homework, not just the completeness of the homework.

The attitudes required of parents and teachers are also clear: neither demonize nor romanticize AI. The statement, 'My child uses AI well,' could mean, 'My child cannot formulate questions independently.' Conversely, completely preventing children from touching AI is also not balance. Both the uninitiated and the dependent are, as Professor Shin diagnosed, in a state of imbalance. At home, spending time reading long texts together and discussing points where interpretations differ is more valuable than an hour of short-form content. In schools, assignments that presuppose AI use and assignments performed without AI should be intentionally mixed. The literacy of handling tools and the literacy of thinking without tools are not the same ability.

The decline of David Deming's 'smart loner' is already beginning. As AI increases individual productivity, the scarcity value of the ability to conflict, persuade, and collaborate with others increases. Literacy is the language of that collaboration. If you cannot read, you cannot read the world of others; if you cannot write, you cannot present your own world. A professor's warning that outsourcing everything for convenience can lead to losing the essence of life is not romanticism, but closer to the common sense of cognitive science. The experience of thinking outside the box, trying and failing, reading books, and writing is life itself.

Learning in the age of AI should be slower. This may sound paradoxical, but in an era where speed is the domain of tools, speed is not the human domain. It is about pausing, rereading, rephrasing in one's own words, and enduring others' rebuttals. From elementary school to university, the superficial appearance of homework can be smoothed by AI. However, if the questions disappear beneath that smoothness, literacy collapses not in scores, but first within a person's inner self. The answer to the question of whether it can be maintained is neither pessimism nor optimism. It depends on whether we cultivate the habit of reclaiming intellectual agency or surrender to convenience. That habit begins not with grand educational reforms, but with 15 minutes of silence before opening today's homework.