Kwon Ki-seong, CEO of Freewillin, selected ‘mathematics’ because the core of edutech.
He began his explanation by saying, “The education market is all about content accuracy and reliability,” and “It can be crucial to offer quality content no matter whether cutting-edge technology is used.”
Freewillin is an AI-based edutech solution and content company founded in 2017. It has gathered 700,000 pieces of its own digital content. Specifically, it uses AI within the production process to extend work efficiency and has its own original works through direct problem creation.
“AI can quickly separate problems, units, and levels of difficulty, and may shorten work time by updating terms from previous courses to the most recent version,” said Kwon. “Nevertheless, I feel AI still only guarantees about 70% accuracy, so I’m only using it for modifications and easy tasks.”
He emphasized that accuracy is a worth that can’t be given up within the education market. Within the case of general content suggestion platforms, satisfaction or trust doesn’t drop significantly even when content that doesn’t match the user’s taste is chosen in keeping with the algorithm. Nevertheless, this isn’t the case with the problematic content of edutech. Even when an error occurs in just one problem out of ten, it has a big impact on trust.
“From this attitude, the applying of AI doesn’t necessarily increase the usability,” he said. “As a substitute, when rule-based systems are used, satisfaction and stability could also be higher,” he explained. As paper is being replaced by electronic devices, the applying of IT technology in education is important, however the tempo and timing are key.
Ultimately, the query is whether or not consumers will accept AI content and services, he said.
Nevertheless, he said that the introduction of AI can’t be delayed due to “equality of educational opportunity.”
The intention is to offer quality education to all students through ‘AI teachers’. The truth is that not all students receive equal education as a consequence of differences in region, economic power, and skill. Also, even under the identical conditions, the outcomes may vary depending on the training style.
In other words, personalized learning is inconceivable.
Data evaluation is the strategy to solve this. For those who accumulate data on the training forms of quite a few students, AI can produce ‘patterns’. It might probably recommend the optimal learning method for every student.
“That’s why math generally is a core subject,” Kwon emphasized.
The Korean language is somewhat weak by way of clarity between ‘higher and lower concepts’. Simply because you solved a 4-point query on a mock test, there is no such thing as a guarantee that you’re going to get a 3-point query right.
Nevertheless, mathematics is comparatively easy to differentiate. It’s because the range of problems that may be solved is revealed depending on the training progress and whether the concept is mastered. Combining these characteristics of mathematics with the information strength of AI makes ‘personal problem recommendations’ possible.
Freewillin also revealed that it has developed an AI-based problem suggestion product with this possibility in mind. It’s the results of having internal problem (content) production personnel, collaborating with publishers, and specializing in engineering. It’s currently searching for technological advancement and an appropriate launch time.
Now we’re specializing in applying AI edtech to universities.
Last June, we launched the AI university education courseware ‘Pulicampus’. That is the results of developing and applying software for basic academic ability assessment with Konkuk University three years ago.
On the time, Konkuk University beneficial it to other universities, including Soongsil University. For example of how easy it was to make use of and the way satisfied it was, it was reported that the satisfaction level of scholars and school was very high.
When a student who had been consistently making mistakes on a particular sort of problem in a subject was allowed to proceed solving AI-generated problems, his rating improved from 66 to 92 points. He said there was quite a lot of positive feedback from students.
In this manner, there are two major points within the university education curriculum. ▲Domestic higher education specialized ‘diagnostic software’ that diagnoses the fundamental academic skills that a university student must have ▲’AI courseware’ that analyzes students’ familiarity and understanding of university classes, etc. The reason is that demand is increasing as non-major admission and integration of liberal arts and sciences are emerging as trends.
CEO Kwon Ki-seong said, “Up to now, we provided college students with online lecture IDs and other support, but that method was not enough to resolve the shortage of actual content.” He identified, “Statistics, for instance, is a compulsory subject in all science and engineering colleges, but there is sort of no diagnostic content or data.”
“I feel we’re setting the usual because there’s significantly less learning diagnostic content in comparison with elementary, middle, and high schools,” he said. “Any further, we is not going to usher in overseas academic ability diagnostic systems, but will work to export domestic math content.”
Meanwhile, you possibly can get vivid feedback from students on the YouTube Freewillin channel.
Reporter Jang Se-min semim99@aitimes.com
