“Grades are proportional to the amount of money invested.” Although this theory has not been verified, it is widely accepted as fact among many students and parents. This is the backdrop against which private education companies, represented by Daechi-dong academies, are generating huge profits. Private education expenses in Korea have risen for four consecutive years, exceeding 29 trillion won last year. To solve the structural problem of ‘rich fathers producing rich sons,’ public education must become strong enough to compete. However, there are numerous tasks that need to be addressed, starting with increasing the number of teachers. Park Chan-yong, CEO of I Hate Flying Bugs, who has been developing customized learning technology for 14 years, says that AI can be a realistic solution to bring about qualitative changes in public education. We spoke with him.
-The phrase “solving the bug of educational inequality” on your company’s homepage is striking. There is a lot of criticism of Korean education as “education for the wealthy, no education for the poor.” Did this problem awareness serve as the starting point?
▶Our company’s most important mission is “equal access to high-quality education.” While most edtech companies aim for “equality of educational opportunities” or “equality in education,” we use a clearer keyword. We emphasize “high-quality” education rather than just “education.” We added the word “opportunity” before “equality” because we believe that universal equality of educational opportunities has already been largely achieved in Korean society. In fact, school education is provided free of charge. Private education services like MegaStudy have also become more accessible, allowing many students to use them at affordable prices.
-So what does “high-quality education” mean?
▶It refers to education that involves direct human management. Education that involves direct human management, like offline tutoring or private lessons, has a labor-intensive structure, and the costs are reflected in the service prices, making them very expensive.
This is not to say that cheaper is better. To achieve “equal access to high-quality education,” the issue is not cost-effectiveness but rather the ability to deliver the same quality as high-cost services costing millions of won at a slightly lower level than the existing 300,000 won education fee. This is necessary to ensure business viability, and as the scale expands, service prices can be further reduced based on profitability. If tutoring achieves 100% results at a price of 500,000 won, consumers will never choose a service that achieves 80% results at 100,000 won, even if the education company creates it.
-What approach are you using to achieve that goal? Recently, many students are using generative AI. Is that a better method?
▶Since there are differences in learning ability among students, a customized curriculum tailored to these differences is most important. Some students may only need 10 training sessions, while others may require over 20 repetitions. Traditional education has not adequately addressed such individual differences. We are now addressing these aspects, enabling us to achieve significantly better results in the customized learning domain. I don’t think using generative AI like ChatGPT to solve problems and get answers is very meaningful, as the solutions are already provided when you purchase a problem book.
-What specific systems do you have in place?
▶We are not just a service that displays learning results on a screen and ends there. We have a structure in place where teachers manage students based on data. Through real-time data analysis, we provide customized learning for each student, and we link this data to an LLM (large language model) so that it can make judgments and intervene like a teacher. AI can intervene in the same way that a teacher would. In simple terms, it initiates communication with the student first. Currently, we hold over 2 billion learning interaction data points and have a platform capable of actually implementing this. This is a key differentiator from general LLM development companies.
-Does this mean AI replaces teachers?
▶It does not fully replace teachers. Humans must guide students based on will and perspective, so the human role remains essential. However, while a single teacher could previously manage only 20 students, using AI agents allows for managing 50 or even 100 students in a high-quality manner. And it’s much higher quality. It provides a higher-quality service at a much lower cost than tutoring centers. If such technologies are widely adopted in public education, students can receive sufficient guidance at school without needing to attend tutoring centers.
-K Edu is also gaining attention overseas. How are you setting your overseas strategy?
▶We need to approach each country differently. In developing countries, addressing basic literacy issues is the top priority. Our exam-focused education system may be suitable for them. On the other hand, advanced country-type services require a different approach. We have solutions that can address both.
– What do you think are the qualities needed in the AI era?
▶From the perspective of nurturing future talent, it is extremely important to cultivate people who can utilize AI well. This may even take priority over receiving education through AI. People who are good at telling AI what to do and getting it to do it will have a competitive edge. However, this is much more difficult than it sounds. Identifying what one desires is a laborious process. Therefore, articulating one’s desires in concrete terms is extremely challenging. Ultimately, in the future society, self-reflection and expression skills, particularly language education, will become more important than anything else.
– How can such skills be cultivated?
▶More important than reading is writing and critical thinking training. The process of clearly identifying what one desires, accurately communicating it to AI, and generating results is crucial. However, current students lack training in articulating what they want. This is because the current curriculum focuses on repeatedly solving predefined problems. In the future, the ability to define problems independently and delegate their resolution to AI will become crucial. In the past, this role was reserved for those in power. Now, it is an era where all citizens must possess this capability.
Yoon Seok-jin, Reporter, Money Today Broadcasting MTN
