Edtech
EdTech (Education Technology)
What is edtech?
- Using technology to transform how learning is designed, delivered, and assessed
- Generative AI speeds up content production and personalization
- Adoption hinges on defining the problem, not the technology
What edtech really changes is not the shape of the content but the observability of learning. In classroom training, knowing who got stuck and where depended on the instructor’s instinct. In a digital environment, logs record the minute of a video where learners dropped off and how many times they missed each question. So the training manager’s job shifts from 'scheduling courses and checking attendance' to reading the data and redesigning learning paths.
Choosing is easier if you sort tools into three layers: an LMS that manages completion records and mandatory training; an LXP that lets learners find content themselves through recommendations and search; and an LRS that collects learning behavior data in standards such as xAPI. Adaptive learning only works once the third layer is in place, and until a certain volume of question-response data builds up, a recommendation engine just repeats popular content. Expecting personalization from a few hundred log entries at a 300-employee company is a sample-size problem, not a technology problem.
Generative AI has shifted the cost curve of this model. Producing one hour of e-learning through an outside vendor used to take weeks from order to delivery. Now, by summarizing existing course materials and lecture recordings and turning them into quizzes, a team can produce a first draft in-house in a few days. The bottleneck, however, has moved from production to review and keeping content current. In areas like laws and safety regulations, where errors lead to accidents, you need to build a human final-approval step right into the content pipeline.
The most common mistake in practice is buying the tool first and finding the problem later. Install a platform that costs tens of millions of won a year, upload your existing PDFs as they are, and six months later monthly logins fall to single digits. Completion rate is not proof of learning. A video can be left playing while someone does other work, and the course still shows as completed. Also, if the contract does not spell out ownership of learning data and the terms for transferring it, the moment you switch vendors, years of learning history disappear.
Seokhyun YoonAI video production and live streaming expert
John KimJohn Kim, CEO of Nanum Consulting and author of B2B Marketing DesignEfforts to bring technology into education date back to audiovisual instruction and computer-based training (CBT), and grew into the e-learning industry with the spread of the internet. Through mobile and cloud in the 2010s and generative AI in the 2020s, 'edtech' has come to mean digital transformation across the whole education industry.
Let’s walk through a hypothetical case. Company A, a Korean auto parts manufacturer with 1,200 employees, had been running statutory mandatory training and job training through an LMS. But annual completion stayed at 43%, and it was an open secret among shop-floor staff that 'you wait for the deadline text, then binge it all at once.' The HR development team had to come up with an improvement plan within three months, with no budget increase.
- When the HR development lead pulled the LMS logs, the average drop-off point in a 42-minute video was around the 9-minute mark, and 68% of enrollments were bunched up in the last three days before the deadline. "They aren’t watching it, they’re just letting it play," someone said in the meeting.
- The training planning team set up a 3-month pilot for 180 production technicians, split each 42-minute video into six 7-minute modules, and sent mobile push notifications twice a week.
- For content, they used generative AI to summarize existing course materials and turn them into quizzes, cutting production from three weeks per module to four days. All 120 safety-regulation questions were released only after a deputy manager on the safety and health team reviewed every one.
- At week 8, a 3-question survey on on-the-job application went out to 34 team leaders. Nineteen of them answered that they had "revised their work checklists based on the training content."
- By the end of the pilot, completion rose from 43% to 81%, and the share of learners cramming just before the deadline fell from 68% to 29%. When the Management Support Division approved a company-wide rollout, it added a learning-data ownership clause to the renewal contract.