[Lim Woong] The real challenge for AI in schools

When large language models such as ChatGPT first entered public conversation a few years ago, most schools treated artificial intelligence as a special-topic event. It stirred fear, curiosity and excitement, but it still felt far away. Guest speakers, mostly tech executives, gave talks about future careers, robots replacing factory work, or the rosy promise that technology would change everything. The message was usually visionary. Teachers listened, took notes, and then went back to school routines that seemed mostly untouched.

ChatGPT, especially from GPT-3.5 onward, and its rivals changed that very quickly. Once students could type in a few commands and get coherent essays, plausible answers to all kinds of questions, or step-by-step solutions to fairly complex math problems, the ground shifted. Skills that schools had spent decades teaching, such as reading, writing and problem solving, no longer looked as sacred as they once did.

Teachers noticed it almost at once. Assignments that used to show real differences in effort and understanding began to produce strangely similar results. So, naturally, many schools tried to ban the tools. Districts blocked access and treated AI use as cheating. But that position did not last, and it was never likely to. Students were already using these tools at home, on their phones and anywhere else adults were not looking.

If an algorithm can produce a competent five-paragraph essay in seconds, what exactly should students be asked to do now? One answer is simple: They should be asked to do harder things. They can critique the generated text, catch factual errors, or question assumptions. They can revise the output and state clearly what they changed and why. They can defend their reasoning aloud. The point is no longer just getting words onto a page. The point is making students “show” judgment. That is where the educational value now lies. The conversation has therefore begun to shift from detection to integration. The issue is no longer merely whether students will use AI. The more difficult question is what kinds of thinking schools should require if students are to retain their sense of agency rather than drift into dependency.

Three domains of AI education now stand out as schools reconsider their curriculum: teaching students to code, preparing future AI specialists and building general AI literacy. Each carries its own risks and rewards.

First, coding education makes sense when it is treated as practice in giving instructions to a machine and understanding how humans interact with computational systems. It can build logic and procedural thinking. But there is a limit to how early and how aggressively this should be pushed. If coding simply becomes another high-stakes subject, schools may end up losing time that younger students still need for reading, writing and basic numeracy. That would be a mistake.

Second, training a future cadre of developers and data scientists is essential for any country that wants to shape AI rather than merely import it. The phrase “sovereign AI” may sound fashionable, but the issue is real enough. Countries that cannot train a high-tech workforce will rely on systems built elsewhere, under assumptions they did not choose. Even so, this track should not be romanticized. Serious AI development requires advanced mathematics, strong scientific preparation and computing resources that most secondary schools cannot realistically provide. It also tends to go wrong when policymakers try to force it too early. Students often do better later, in college and graduate study, when they have first built a strong foundation and had time to develop interests and mature intellectually.

Third, broad AI literacy is now indispensable, but it has to be defined carefully. It cannot mean a few catchy lessons about chatbots. It also cannot mean a stripped-down technical survey that only the mathematically confident can follow. Students need history, civics, literature and the sciences if they are to understand what these systems mean in real-world contexts of bias, privacy, labor, equity and power.

What we need, then, is not a single pathway but two. Most students need a solid general education in what AI is, how it works at a basic level, where it is useful and where it is dangerous. A smaller group needs deeper preparation in mathematics and science. Those are two different goals. A sensible system should be able to pursue both without confusing one for the other.

At the same time, AI is not just adding one more topic to the curriculum. It is changing the shape of work in many fields. In the life sciences, protein-folding prediction has already changed parts of the research process. In law and policy, document retrieval and pattern recognition are altering how professionals search, compare and make arguments. Robotics, helped along by new chips for perception and control, is moving into hospitals, farms and homes, not just factories and warehouses.

These changes bring real challenges with them. The hardest part will be implementation. School systems have a habit of taking a new idea and immediately asking how to “test” it. That temptation is already visible in discussions of coding standards, AI ethics requirements and possible exam content. Bureaucracies like measurable things. But forcing AI education too quickly into an exam-centered framework would create trouble from the start. Existing subjects already compete for limited time, and any new requirement will trigger pushback. Some of that resistance will come from self-interest, certainly. Some of it will come from a reasonable concern that schools are piling new demands onto old structures without thinking hard enough about coherence.

Schools now face a rare opportunity to decide what students should still learn to do for themselves in a world where machines can handle much of the routine work. Real progress will require moving away from the exam-driven selection system that has long defined what counts as talent. Only then can schools cultivate young talent in science and technology, while also preparing the next generation of citizens who can think clearly and ethically about what these systems should and should not do.

Lim Woong

Lim Woong is a professor at the Graduate School of Education at Yonsei University in Seoul. The views expressed here are the writer’s own. — Ed.

khnews@heraldcorp.com