What Happens When AI Is Free ... South Korea Tried It

A 12-minute explainer from Economics Explained, uploaded on 14 October 2026, on South Korea’s promise of free AI for everyone, announced about a month earlier. The narrator’s case is that it is more modest than the “universal basic compute” headlines. It has two parts. One is a government-services assistant wired into state databases (passports, welfare, tax), which he calls a customer-service bot that replaces waits for junior clerks. The other is free everyday chat, delivered through three domestic firms (Kakao, KT and SK Telecom), reachable through KakaoTalk and also ordinary phone calls and texts for people who struggle with apps. The government supplies GPU capacity and modest payments, and at least half the use has to run on Korean foundation models (with more from other domestic developers), which in effect pays local firms to take market share from US and Chinese rivals. The models are basic; the bet is lock-in through memory and attachment, with charges once the trial ends.

He weighs two precedents. Korea’s national broadband push seeded its tech sector; El Salvador’s national Bitcoin wallet fell far short of the hype. He thinks AI is likelier to deliver real value than a speculative asset. The case for it is demographic: with fertility far below replacement, Korea is a preview of other economies, and making fewer workers more productive is the intuitive answer. The case against is the bottom rung. Young Koreans already struggle to find entry-level work, and automating it could make that, and the birth rate, worse. He also notes the timing: it was planned at the height of an AI stock mania the government fed rather than cooled. Folding public services into private platforms with subsidised lock-in, he concludes, fits the techno-feudalism the channel has covered before.