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

Summary

South Korea is piloting free AI access for citizens in response to a declining labor force and as a way to develop domestic AI companies. The initiative could raise productivity and encourage entrepreneurship, but it also carries fiscal, employment, and market-concentration risks.

The video’s central question is whether the program will become productive public infrastructure—like Korea’s earlier broadband investment—or an expensive technology bet with limited adoption.

How the program works

• Government-services AI: A system connected to government databases would help people with tasks such as passport applications, social-benefit applications, and tax questions. It is intended to handle routine queries that might otherwise go to entry-level civil servants.
• General-purpose AI access: Citizens can use basic AI chatbots through services from Kakao, KT, and SK Telecom. Access through KakaoTalk, phone calls, and text messages is designed to make the service easier to use, including for older people.
• Infrastructure and funding: The government is supplying GPU server capacity—reportedly hundreds of advanced Nvidia server nodes—and making small annual payments during the trials. These payments are not expected to cover the full operating cost.
• Domestic-market condition: About 80% of traffic must go through Korean AI models. The program therefore also functions as a subsidy and customer-acquisition channel for domestic providers, helping them compete with paid US and Chinese alternatives.
• Limited service tier: The offerings are described as closer to unlimited free-tier AI than to expensive, advanced models. Providers may benefit if users develop habits and relationships with the services, then later pay for premium features or continued access when government support ends.

Business and policy approaches discussed

• Public technology investment loop: Make infrastructure broadly available, encourage businesses and individuals to find useful applications, raise productivity and economic activity, and generate tax revenue that may offset the original investment.
• Domestic ecosystem-building: Use public infrastructure and subsidies to create demand for local providers, build technical capability, and establish domestic products before foreign alternatives dominate.
• Two-track deployment: Apply AI to both routine public-service operations and general-purpose access, while treating them as distinct use cases with different users and success measures.
• Adoption and retention strategy: Free access can reduce barriers to trial and build user habits. Providers may later convert some users to paid tiers. This potential commercial benefit is an incentive for participating tech companies, even if the trial itself is not profitable.
• Technology bets need outcome tests: Korea’s broadband rollout is presented as a successful precedent, while El Salvador’s national Bitcoin wallet is presented as a cautionary example. The distinction is whether the technology produces broad, practical productivity gains and real adoption, rather than hype or speculative value.

Potential benefits and risks

• AI could help workers do more with fewer people as the population ages and the workforce shrinks, particularly in intellectual work and manufacturing.
• Automation may also reduce entry-level opportunities. If young people cannot find stable work, that could further weaken their incentives to start families, worsening the demographic issue the AI program is meant to help address.
• The policy could strengthen Korean AI firms, but it could also create dependence on particular corporate platforms and blur the boundary between public infrastructure and private profit.
• The timing may fuel AI-related investment hype. The video notes that the initiative was developed during a period of strong AI enthusiasm and announced amid a stock-market decline and broader uncertainty.
• The program’s long-term value depends on whether it creates measurable productivity and domestic capability—not simply high usage or a large user base.

Metrics and figures cited

• Population: South Korea has about 51 million residents.
• AI allocation: Roughly 80% of program traffic is to pass through Korean AI models.
• Infrastructure: The government is reportedly providing hundreds of advanced Nvidia server nodes; the transcript gives no total cost or detailed budget.
• Demographics: The fertility rate was 0.72 in 2023, with preliminary 2024 data suggesting about 0.68.
• Labor market: Youth disengagement—defined in the video to include discouraged workers and involuntary part-time employment—is around 18%.
• Shopify-sponsored claims: The video says entrepreneurship added more than $850 million to South Korea’s GDP over four years. It also cites Shopify-platform economic activity of $1.2 trillion globally since 2022, more than double its previous level; $687 billion in US economic activity “this year”; and more than 5.6 million new business registrations last year. These figures are presented in the sponsor segment, not as measured outcomes of Korea’s AI program.

Practical evaluation points

The video does not provide formal targets for the trial. Useful measures for assessing the initiative would include:

• Public-service resolution time and cost
• Worker productivity
• Usage and repeat adoption
• Fiscal cost per active user
• The share of activity handled by Korean models
• Whether AI adoption creates or displaces entry-level jobs

These measures could help distinguish durable productivity gains from subsidized usage or hype.

Presenters and sources mentioned

• Economics Explained: Video narrator and publisher.
• South Korean government: Source of the AI-access initiative described.
• Kakao, KT, and SK Telecom: Companies operating the general-purpose AI services.
• Shopify: Sponsor; its platform and entrepreneurship figures are cited in the sponsor segment. The video also mentions Shopify’s AI assistant, Sidekick.
• Economics Explained newsletter: Promoted in the video.