The widely heralded promise of explosive AI-driven growth is not a simple, inevitable trajectory, but rather a complex economic and societal challenge fraught with critical bottlenecks and distributional pitfalls. Professor Luis Garicano, a distinguished economist from the London School of Economics and former EU parliamentarian, recently engaged in a sharp discussion with Epoch AI's Anson Ho and the University of Edinburgh's Andrei Potlogea. Their conversation, titled "The EU and the not-so-simple macroeconomics of AI," critiqued prevailing optimistic views and highlighted significant policy and labor market concerns, grounded in Garicano's unusual institutional lens.
Garicano immediately challenged the notion of unfettered, explosive growth, pointing to the often-overlooked institutional and organizational friction that slows technological adoption. He emphasized that even the most advanced AI innovations face significant real-world hurdles, from regulatory approvals—like those required by the FDA for biotech advancements—to the sheer human effort needed for widespread learning and integration. "As long as the AI needs your supervision, because it makes lots of mistakes, then the bottleneck is the human," Garicano stated, underscoring the enduring human element in the current stage of AI development. This human bottleneck, he argues, is often underestimated by those observing AI's rapid progress solely through the lens of coding and research.
A central theme of Garicano's analysis is the question of agency in shaping AI's future. He critically examines the optimistic assertion that "we can direct technology," questioning who "we" actually represents in a globalized, multi-stakeholder environment. Is it China, the US, powerful corporations, workers, or even truck drivers? These diverse interests, often at odds, make a unified, benevolent direction of technological change highly improbable, leading instead to a complex interplay of competing forces.
The economic landscape shaped by AI is also prone to a "superstar effect," concentrating immense value at the top. Garicano highlights how AI acts as a powerful lever, amplifying the productivity and market reach of already highly skilled individuals. "A very good AI programmer with lots of AI can have enormous leverage and can reach a very large market size," he explained, suggesting that this dynamic will exacerbate existing inequalities as a select few capture disproportionate gains. This phenomenon contrasts sharply with the broader labor market, where AI's impact is far less benign.
Indeed, Garicano expressed deep concern about the "missing training ladder" for entry-level workers. While non-autonomous AI can initially complement human workers, helping juniors perform routine tasks more efficiently, this very efficiency erodes the traditional pathways through which new entrants gain experience and climb the skill hierarchy. If basic tasks are increasingly automated, future workers lack the foundational "menial tasks" that historically served as on-the-job training, leading to a societal market failure. This is often not about direct job displacement, but a reduction in hiring for junior roles. "It seems like a lot of people are not hiring junior employees," Garicano observed, based on recent aggregate data.
The EU's approach to AI regulation also drew sharp criticism. Garicano, drawing on his experience in the EU parliament, views the EU AI Act as potentially stifling innovation rather than fostering it. He drew parallels to the GDPR, arguing that such regulations, while well-intentioned, can inadvertently create competitive disadvantages. "Part of the risk is you try to control the technology and you end up without technology," he warned, suggesting that over-regulation could lead to Europe falling behind in the global AI race.
In the short run, AI's macroeconomic effects are likely to be characterized by significant labor and capital reallocation. If AI renders certain sectors or tasks effectively "free," the displaced workforce and capital must find new applications. This transition period could lead to unemployment, de-skilling, and a painful adjustment for many workers whose human capital is suddenly devalued. The challenges of retraining and redeploying a large workforce are immense and often underestimated.
AI's impact isn't uniformly beneficial. It creates significant short-term dislocations.
Ultimately, Garicano’s commentary offers a vital, grounded perspective on the complex macroeconomic implications of AI. He urges policymakers and industry leaders to move beyond simplistic narratives of inevitable progress and confront the intricate challenges of institutional inertia, distributional inequity, and the profound transformation of the labor market. His insights compel a more cautious, yet proactive, approach to shaping AI's integration into society.
