In a world buzzing with promises of an AI-driven revolution, one of the world's top economists is sounding a note of caution. The expected productivity boom from AI might not be the rocket ship many believe it to be, and the reasons lie in reality.
A Dose of Economic Reality
The
cautionary voice belongs to Torsten Sløk, chief economist at Apollo Global Management, a major global asset management firm. In recent analyses, Sløk has argued that while artificial intelligence is undoubtedly a transformative technology, its path to delivering a broad economic 'payoff' will be much slower and more complicated than the market hype suggests. His argument isn't that AI is a fad; it's that the journey from a powerful new tool to a tangible, economy-wide productivity miracle is long and filled with friction. This perspective stands in stark contrast to the narrative of immediate disruption that has propelled tech stocks to staggering heights, suggesting that investors and business leaders might be getting ahead of themselves.
The Irreplaceable Role of Judgement
A central pillar of Sløk's argument is the distinction between data processing and true business judgement. Current AI, particularly generative AI, is exceptionally good at summarizing information, writing code, and identifying patterns in vast datasets. What it cannot do, however, is replicate the nuanced, context-aware decision-making that defines senior leadership. Business judgement involves weighing risks, understanding office politics, interpreting subtle market signals, and making strategic bets based on incomplete information and intuition built over decades. An AI can present three potential marketing strategies based on data, but it can't decide which one best aligns with the company's long-term brand identity or how a key competitor is likely to react. This 'judgement gap' acts as a natural brake on AI's ability to automate the most valuable corporate functions.
Beyond the Hype: AI's Practical Hurdles
The limits of AI extend beyond the boardroom. Sløk points to several practical hurdles that stand in the way of rapid, widespread adoption. Firstly, the cost of implementation is enormous, not just for the technology itself but also for the specialized talent required to deploy and manage it. Secondly, integrating AI into legacy systems and complex corporate workflows is a significant challenge fraught with technical and operational difficulties. There are also critical issues around data privacy, security, and the risk of 'hallucinations' where AI models produce confident but incorrect information. Unlike a simple software update, deploying AI effectively requires a fundamental reimagining of business processes, which takes time, capital, and a level of organizational change that most companies are not yet prepared for.
Echoes of Past Tech Booms
History offers a useful guide. Sløk's caution is reminiscent of the 'productivity paradox' observed during the early days of the computer and internet revolutions. In the 1980s and 1990s, economists were puzzled that massive investments in information technology weren't showing up in national productivity statistics. The benefits, it turned out, took decades to fully materialize as businesses learned how to reorganize themselves around the new technology. The same was true for electricity. It wasn't just about plugging in machines; it was about redesigning entire factories and supply chains. The lesson is that transformative technologies rarely produce immediate economic gains. There is a long and often frustrating lag between invention and widespread economic impact, and Sløk’s analysis suggests AI will be no different.
