In the second part of a conversation between Nobel laureates Simon Johnson and Philippe Aghion, the economists reviewed Europe’s technological lag, industrial policy options and the societal risks of artificial intelligence. They argued that recent geopolitical shocks — including Russia’s invasion of Ukraine and Donald Trump’s attacks on the post‑World War II order — exposed Europe’s vulnerabilities, but the US–China rivalry over AI has made the continent’s innovation shortfall especially apparent.
Why AI and industrial policy matter for Europe
Philippe Aghion stressed that Europe’s challenge is not only fiscal or political: while the euro is a strength, it needs to be paired with stronger technological performance to translate into influence. He noted that Europe is often criticized as a “regulatory giant and budgetary dwarf,” and that excessive Brussels regulation can constrain member states’ room for manoeuvre. At the same time, Aghion argued that EU competition policy must become more innovation‑oriented and incorporate industrial policy in key areas such as AI, biotechnology, defense and the energy transition.
Aghion believes Europe should both regulate and support AI: regulation should avoid unnecessarily blocking new entrants, while Europe must build AI infrastructure (computing capacity and data‑sharing) where shortfalls exist. Competition policy and DARPA‑type industrial programs can complement each other to foster competition and help new European unicorns emerge.
The debate on AI’s growth impact: how large is the potential?
Aghion and Daron Acemoglu differ on AI’s macroeconomic impact. Aghion and co‑authors — with Simon Bunelle — estimated roughly a 0.7 percentage‑point annual boost to total factor productivity from automating tasks related to producing goods and services, whereas Acemoglu’s estimate is about 0.07 percentage points. Thus the disagreement is partly empirical.
Aghion emphasized that AI not only automates tasks but also facilitates idea generation and recombination. Ongoing research in Paris suggests that this innovation‑enhancing effect could be a durable source of growth: Aghion estimates it could add at least 0.2 percentage points to annual growth, though full quantification is still in progress.
Competition risks and market entrenchment
Aghion warned that failure to sustain competition could replicate the IT revolution’s pattern: rapid growth followed by the rise of superstar firms — such as Google, Microsoft and Amazon — that become so dominant they inhibit new entrants. Johnson noted that the EU’s multi‑actor governance could be an advantage in crafting competition‑friendly industrial policy, provided EU rules remain flexible and supportive of innovation.
Lessons from China
Aghion acknowledged that China has successfully combined state industrial policy with a degree of competitive pressure, for example through yardstick competition among regions and firms. Prior research he co‑authored found that industrial policy was more effective when multiple firms in the same sector received support — suggesting that competition‑friendly deployment of support matters.
Yet Aghion cautioned that China’s model is top‑down and focused on capital accumulation and automation; it remains uncertain whether it will produce enough good jobs to sustain social satisfaction and long‑term growth. Given China’s demographic decline, AI’s expansion may partially offset labor shortages.
Education: why Aghion advocates AI‑free schools
Perhaps the most concrete and controversial recommendation from Aghion was to exclude AI from schools. His argument is that students must first learn to read, write and solve mathematics independently; introducing AI too early and too pervasively risks undermining autonomous thinking and could lead to a generation that is intellectually atrophied. Practical elements of his proposal include doing homework at school, ensuring high‑quality teachers, small class sizes and AI‑free learning phases.
Aghion also stressed that education is critical for a country’s “absorptive capacity” — its ability to adopt and benefit from new technologies — which is especially important for developing countries.
Conclusions: opportunities paired with risks
Aghion is optimistic about AI’s long‑run growth potential but cautions about the risks from weak competition and adverse political reactions. Europe’s tasks are to build competition‑friendly industrial policy and AI infrastructure while protecting core educational skills through periods without AI. Much depends on empirical answers — especially regarding the magnitude of AI’s productivity gains — but policymakers must address regulation, competition and education together if they want to realize AI’s benefits without sacrificing social cohesion.
Philippe Aghion is a professor at Collège de France and the London School of Economics, whose research focuses on the economics of growth; he received the Nobel Prize in 2025. Simon Johnson is a professor at MIT, former chief economist of the IMF, and the UK’s AI envoy; he received the Nobel Prize in 2024. This interview was published in two parts; this is the second, focused on AI and education.
