Why it matters. The debate over AI and jobs is global, and Frey's argument reframes it as a policy choice rather than an inevitable tech outcome — a lens that applies to any country bracing for automation.
Background. Hankyoreh is a center-left South Korean daily known for labor and social-justice coverage; its annual Human and Digital Forum is a prominent venue for tech-and-society debate. Korea faces one of the world's lowest birth rates and fastest-aging populations, and its unionized manufacturing workers — especially in autos and shipbuilding — have a long history of resisting automation, which is why robotics adoption is politically charged there.
What to watch next. Watch whether Korea pairs its aggressive push into robotics and 'physical AI' with the kind of worker safety nets Frey credits for the Nordics' calmer response.
Choice, Not Technology, Will Decide the Future of Work
As layoffs ripple through Big Tech, Oxford economist Carl Benedikt Frey argues that the impact of automation hinges on political and social choices rather than technology alone — and he points to the Nordic “flexicurity” model, which pairs easy hiring and firing with strong safety nets, as a realistic way to ease public anxiety about artificial intelligence. Frey made the comments in a video interview ahead of his keynote at South Korea’s fifth Hankyoreh Human and Digital Forum on June 24, where the theme is the future of work and learning in the AI age.
Frey is best known for a 2013 study, co-authored with Michael Osborne, that warned 47% of U.S. jobs were exposed to automation. He stresses the report measured exposure, not predicted losses, and that the three bottlenecks it identified — complex social interaction, creativity, and coping with unpredictable situations — still hold. Generative AI, he says, recombines existing concepts rather than inventing genuinely new ones, so humans remain necessary to verify its output. “Training a model on Impressionist paintings won’t produce conceptual art,” he notes.
Why Workers Feel Calmer in the Nordics
Frey cautions that AI’s labor-market effects differ sharply by country, so one nation’s experience can’t be copied onto another. In the United States, he sees a broadly deteriorating job market for young people — worse in manufacturing than in white-collar work — driven partly by AI but also by capital rushing into AI, which raises borrowing costs and hurdle rates elsewhere and pushes firms toward layoffs.
The biggest risk, he argues, falls on young people and new labor-market entrants. If their hiring dries up, the “pipeline” that builds skilled workers could collapse. His proposed fix is time: give job-seekers enough support to find a good match. Sweden’s flexicurity system, he says, lets workers lose a job yet maintain a decent life while searching, with industry-level employer associations and unions cooperating to avoid chaos. That security, he believes, is exactly why Nordic workers feel far less anxious about automation than others.
Inequality, Robots, and an Aging Korea
Frey is careful not to predict whether AI will widen or narrow inequality. He compares it to how GPS reshaped taxi driving: knowing every street name stopped being a valuable skill, so more people could drive, but existing drivers’ wages fell. Similarly, if AI makes accounting easier, more people can do it and incumbent pay drops. He also expects some jobs in high-cost cities like London, New York, and Seoul to move offshore to places like Manila — possibly widening inequality within nations while narrowing it globally.
On Korea specifically, Frey says its shrinking population — a challenge it shares with Europe — can only be addressed through productivity gains and advanced robotics, including “physical AI” for elder care. He urges against placing excessive barriers on adopting such technology, while insisting the transition be made as inclusive as possible. Yet inclusion, he warns, must not kill “creative destruction”: without technological progress, societies stagnate and ultimately fare worse. “Some disruption is better than stagnation.”
Ultimately, he says, demand matters as much as supply — societies themselves decide which jobs they want done by AI and which by humans.
