This episode exposes the EU's self-imposed regulatory hurdles, arguing they risk stifling AI adoption and exacerbating Europe's demographic and debt crises. Professor Luis Garicano dissects the macroeconomics of AI, revealing why the continent's "smart second mover" strategy might fail without radical policy shifts.
AI's Autonomy Threshold and Economic Disruption
- Human Bottleneck: As long as AI makes mistakes, human supervision remains the limiting factor. AI assists, but human time constrains overall output.
- Productivity Jumps: Autonomous AI, like a fully self-sufficient AI lawyer, creates a "discrete jump" in productivity by removing the human bottleneck entirely.
- Short-Run Macro Shock: The rapid productivity increase in one sector (e.g., medical/legal services becoming "free") necessitates labor and capital reallocation. This can lead to massive human capital depreciation and potential short-run recessionary pressures due to reduced demand.
- Distributional Bifurcation: AI initially boosts junior worker productivity (Brynjolfsson's customer support chatbots). However, aggregate data suggests AI replaces junior roles through reduced hiring (Lichtenberg & Houssini's "seniority-based technological change," Brynjolfsson's "Canaries in the Coal Mine"). Simultaneously, top-tier AI experts experience "superstar effects" (Sherwin Rosen), gaining enormous leverage and market size.
“As long as the AI needs your supervision because it makes lots of mistakes, then the bottleneck is the human.” – Luis Garicano
The AI Becker Problem: A Disappearing Training Ladder
- Devalued Apprenticeship: Apprentices traditionally "pay" for training through menial tasks (e.g., contract review, basic research). AI's ability to perform these tasks cheaply devalues this payment mechanism.
- Training Ladder Disruption: Firms become reluctant to hire and train junior workers if AI can perform their initial tasks more efficiently. This threatens the transfer of tacit knowledge—undocumented, experience-based skills—from experts to new entrants.
- Growth Hole Risk: A disappearing training pipeline creates a long-term "hole" in human capital development. Without new experts, future AI supervision and innovation could falter, impacting economic growth.
- Education's Dual Challenge: AI acts as an "enemy" for foundational learning, as students use it to bypass basic skill acquisition. Conversely, AI enables students to tackle vastly more complex, real-world problems (e.g., trade models, policy analysis) than previously possible, demanding a radical shift in pedagogical approaches.
“If the AI can do the basic research at McKenzie, can do the contract review at Cravath… then how do you pay for your training?” – Luis Garicano
Europe's Regulatory Overreach and Geopolitical Vulnerability
- Uncontrollable Trajectory: Garicano dismisses the idea of "directed technical change" (Daron Acemoglu), citing a game-theoretic race between the US and China. Slowing down is not an option; non-participation means being left behind.
- The "Who is We?" Problem: The notion of "we" directing technology is ambiguous, as firms, workers, and governments have divergent interests.
- EU AI Act's Flaws: The Act, drafted during a "panic" moment (ChatGPT's release), imposes stringent, often impractical, requirements:
- Forbidden Uses: Emotion detection, social scoring.
- High-Risk Applications: Education, health, energy infrastructure require error-free, unbiased data (which doesn't exist), 10-year record-keeping, conformity assessments, and registration with 55 authorities.
- GDPR Precedent: Garicano points to GDPR's documented negative impact on EU venture capital and startups as a warning.
- Risk of No Technology: Europe risks controlling technology so tightly that it ends up without it. The continent lacks competitive foundation models (2 in Europe vs. 50 in the US) and struggles with AI implementation.
- UK's Divergence: The UK, post-Brexit, adopts a more pro-AI stance, leveraging talent hubs like Oxford and Cambridge and companies like DeepMind, positioning itself as a potential "Silicon Valley."
“Part of the risk is you try to control the technology and you end up without technology, which is kind of the world where Europe has a risk of finding itself.” – Luis Garicano
AI Value Chain and Europe's "Smart Second Mover" Strategy
- Hardware Dominance: The hardware layer (chips, GPUs) is dominated by the US and China, with steep learning curves and high value capture. Europe has no entry possibility here.
- Cloud Computing Risks: Cloud providers (US companies) pose geopolitical and economic risks due to switching costs and extraterritorial laws (US Cloud Act). Europe must ensure data encryption, local server presence, and data portability.
- LLM Layer Competition: The foundation model layer (LLMs) currently exhibits strong competition and open-source alternatives (Llama, Mistral), making competitive advantage difficult to sustain unless memory and personalization create switching costs. Regulation must ensure interoperability and data portability, unlike social media.
- Europe's Strategy: A "smart second mover" approach involves "free riding" on US/China investment in LLMs and data centers. Europe should focus its resources on developing a strong implementation layer, ensuring autonomy, local data centers, and robust interoperability.
- Geopolitical Headwinds: The US government actively supports its tech giants, making it difficult for Europe to enforce level playing fields or interoperability. Europe's geopolitical dependence (e.g., for defense) limits its leverage in trade disputes.
- Sovereign Compute: While desirable, public investment in data centers is insufficient given the scale of private sector spending (hundreds of billions annually). Private investment, often by US tech giants, will drive local data center development, offering some local control but not true sovereignty.
“If you don't have the hardware, everything else flows downstream.” – Luis Garicano
Macro-Financial Stability and Demographic Imperatives
- Interest Rate Dynamics (R vs. G): The economic consensus suggests AI will increase both interest rates (R) due to higher capital productivity and investment, and economic growth (G). Fiscal sustainability depends on R-G. If G > R, debt is sustainable.
- Europe's Fiscal Peril: Garicano warns Europe faces the "bad part" of higher global interest rates (R) without the "good part" of higher growth (G), due to AI adoption obstacles (human bottlenecks, regulation). This exacerbates existing high explicit and implicit debt burdens (e.g., pensions).
- Demographic Collapse: Global fertility rates are plummeting, including in developing nations, leading to countries "growing old before they get rich." This creates an urgent need for AI to boost productivity and address labor shortages, particularly in care economies.
- AI in R&D: AI's most significant impact on growth lies in accelerating the production of new ideas and scientific research (Jones, Aghion). Given demographic trends, AI-driven R&D is crucial to offset declining human capital inputs.
“What I worry about Europe is that you are going to have the bad part of having to pay higher rates without having the good part of having higher growth rates.” – Luis Garicano
Investor & Researcher Alpha
- Capital Reallocation & Sectoral Shifts: Investors should anticipate significant capital depreciation in sectors heavily impacted by autonomous AI. Focus shifts to companies facilitating labor reallocation or those leveraging AI for "superstar" human capital. Research into the speed and friction of cross-sector labor mobility is critical.
- Training Tech & Human Capital Solutions: The "AI Becker Problem" signals a looming crisis in human capital development. Investigate startups and research addressing new training paradigms, potentially involving AI-driven personalized learning or novel human capital financing models (e.g., equity-like arrangements, though historically challenging).
- Geopolitical Regulatory Arbitrage: Europe's regulatory environment creates a clear divergence. Researchers should model the economic impact of the EU AI Act on innovation and investment. Investors should prioritize regions with pro-AI postures (e.g., UK, parts of US) for AI infrastructure and foundation model development, while focusing on EU-based AI implementation companies that can navigate complex compliance.
Strategic Conclusion
Europe's current AI regulatory path risks economic stagnation, exacerbating its demographic and debt crises by stifling growth and innovation. The continent must urgently pivot from over-regulation to a bold, pro-growth stance, prioritizing AI adoption and fostering an environment where private capital can drive the necessary technological transformation. The next step is a radical re-evaluation of the EU AI Act and a commitment to competitive, interoperable AI ecosystems.
