What are the implications for Europe’s economy? Expert perspectives.
“The [AI] takeoff has started,” wrote OpenAI CEO Sam Altman recently. With this quote, economist Anton Korinek opened a high-level workshop last Thursday, co-organized by CSH, the Austrian National Bank (OeNB) and the Centre for Economic Policy Research (CEPR).
The “AI takeoff” describes a turning point: AI systems that today write texts or create images could soon independently conduct research, solve complex problems, and even develop better AI systems. Korinek, currently spending a research year at CSH, referred to expert assessments suggesting that we are entering such a phase—where AI systems’ capabilities rapidly advance to the point that they soon outperform humans in most cognitive tasks.
Moreover, robotics is also making huge strides and is rapidly advancing. According to Korinek, the limiting factor in the past may not have been the hardware, but the software of robots. Now that the software has improved significantly, we’re seeing major breakthroughs.
WHAT DOES THIS MEAN FOR EUROPE’S ECONOMY?
And how can Europe prepare for it? These questions were at the center of the workshop, which featured leading international experts—including Skype co-founder Jaan Tallinn and AI policy expert Daniel Susskind.
AI‘s Promise: Unprecedented Opportunities
The potential benefits of AI are enormous. Conservative estimates suggest AI could boost productivity by around 3%, though many projections reach significantly higher figures. This mirrors the historical pattern that has made us wealthier: “The reason we’re much more prosperous today than 200 years ago is higher productivity,” explained Anton Korinek, who is also a professor at the University of Virginia and was recently appointed to the Anthropic Economic Advisory Council. Higher productivity creates more output, which generates more capital in a reinforcing feedback loop. Labor performed by AI rather than humans could trigger a similar cycle, dramatically increasing economic output and leading to greater prosperity.
Beyond productivity gains, AI could accelerate scientific progress, enable better work-life balance through reduced working hours, and fundamentally reshape how we approach problem-solving.
The Perils: Significant Risks Ahead
At the same time, the experts warned of major disruptions in the labor market. Daniel Susskind, who has published several books on the topic, explained that unemployment could rise. He outlined how the labor market could face both frictional and structural technological unemployment. Frictional unemployment may result from multiple mismatches: workers’ skills may not align with available jobs, they may not be located where work exists, or they may prefer unemployment over jobs in unfamiliar sectors—men, for instance, might resist working in elderly care. Structural unemployment poses an even greater challenge, as there may simply not be enough work for everyone, potentially leading to increasing socioeconomic and gender inequality.
The development could be particularly drastic in programming: “The leading AI labs are predicting that internally, the last line of code will be written by humans next year,” says Korinek.
For a long time, economists believed that machines wouldn’t be able to perform certain tasks—like writing poems or making medical diagnoses. They were wrong because they assumed that machines would have to think like humans. In other words, they observed how people solve problems and tried to replicate that process. It’s a bit like the early days of aviation, when people thought airplanes would need flapping wings in order to fly.
The same applies to intelligence. The breakthrough came when we stopped assuming that AI has to work like a human mind. Take the medical example: AI diagnoses a condition in a completely different way than a doctor. For instance, it can compare thousands—even hundreds of thousands—of images of moles or cancerous growths, far more than any doctor could see in a lifetime. But AI doesn’t necessarily understand medicine in the way we do.
However, work provides more than income—it often gives meaning and purpose to people’s lives. What happens to society when fewer people work? If AI truly replaces human labor, fundamental questions arise about taxation and social systems: who pays taxes when AI agents generate the wealth? Additional concerns include potential restrictions on freedom, as technology can always be controlled, and the concentration of power among a few tech giants.
Another very immediate challenge was discussed in the Policy Panel: One of the difficulties that companies are currently facing is that expert assessments of future AI capabilities vary dramatically. But they need to make plans and invest now to not be late. Businesses must make decisions under extreme uncertainty, complicated by the fact that corporate decision-making typically takes longer than the currently rapid pace of AI development. It’s like driving from one city to another and having to decide at every intersection whether to turn left or right—both might be possible, but you only find out afterward which route would have been better.
Europe's Challenge
Given this wide range of scenarios for AI’s economic impact, Europe must prepare for all possibilities, including rapid acceleration of AI beyond its current capabilities. Currently, Europe is lagging behind the US and China in AI development, experts warned. They emphasized that Europe should place greater emphasis on recognizing and seizing opportunities to deploy AI rather than training foundation models from scratch, alongside managing risks of AI. While regulatory harmonization would be ideal, what should Europe do if the USA and China surge ahead?
“The most dangerous strategy is to ignore ongoing AI progress—because certain negative consequences will materialize regardless of whether we take advantage of the positive ones,” said Korinek.
While Europe doesn’t necessarily have to lead in developing AI technologies, it could make far better use of existing technology across the entire value chain. As a cautionary example, he pointed to the European automotive industry, which has lost competitiveness in recent years—mainly due to rapid advances in China.
Three Key Government Roles
Korinek identified three key tasks for European governments: enabling AI, using AI, and governing AI—that is, ensuring it aligns with societal goals.
While AI already serves as a powerful tool for cognitive workers—some power users are becoming 2–5 times more productive—most people have not yet realized that AI will fundamentally transform our world. This disconnect between current capabilities and future potential creates unique challenges for decision-making.
CSH President Stefan Thurner emphasized the urgency of the moment: “The AI revolution in science is already well underway—with promising advances, but also significant challenges. It’s clear that similar developments will soon affect society as a whole. We need to be prepared—because the worst thing would be not understanding what is happening right now.”
Martin Summer, Head of the Research Section at the OeNB, expressed optimism: “Even if Europe is not currently leading the race in AI development, we can rely on the strengths of our continent: a highly skilled workforce, robust institutions, and the ability to introduce this revolutionary technology in a thoughtful and responsible way.”
If the “AI takeoff” is happening, we should observe massive resources flowing into AI development, including chips, talent, energy, and funding, alongside rapid advances in AI capabilities. Over time, symptoms would include rapid scientific progress beyond AI fields, the relative devaluation of cognitive labor as intelligence approaches zero cost, a relative increase in the value of physical labor, rapid output expansion, and rising interest rates.
Education and Preparation as Key to Success
The experts agreed that education systems must adapt. AI literacy and fluency needs to be taught at all levels of education—from schools and universities to professional training. What skills will still be needed in an AI-dominated job market? How and when should they be learned? Or do we need to fundamentally rethink what education means in a world where AI surpasses many human abilities? These questions will determine Europe’s future resilience.
Blind optimism and doomsday thinking won’t move Europe forward, the experts concluded. What’s needed instead is sober analysis, strategic action, and a degree of flexibility—because possible future scenarios still differ widely. The AI revolution is happening—with or without Europe. The question is not whether the transformation will come, but how Europe will make it work to its advantage.
The extremely rapid developments we are witnessing may be unprecedented in their speed. Korinek explained that this acceleration started in 2022 with large language models and shifted gear with reasoning models in 2024. These systems generate long chains of thought in the background (sometimes generating content equivalent to hundreds of books) before providing responses that are often deeply insightful—a quantity and speed impossible for humans. “In some way, quantity produces its own form of quality,” he noted.
The next major development in 2025: AI agents—autonomous systems that pursue goals by combining strategic thinking with external tools such as internet search or writing code. “We could soon see companies that consist of just one person and thousands of AI agents and yet are worth billions,” said Korinek.
