Daron Acemoglu, a renowned economist at MIT, wants to set the record straight from the start: he is not an opponent of artificial intelligence. In fact, he recognizes its vast potential. “I’m not an AI pessimist,” he says early in the discussion. However, despite his optimism for AI’s possibilities, Acemoglu is deeply concerned about the current hype driving a significant investment surge, particularly among tech stocks. He believes the frenzy around AI might not lead to the productivity revolution many expect, raising concerns about the financial risks involved.
According to Acemoglu’s analysis, AI will have a limited impact on the labor market over the next decade. He estimates that only 5% of jobs could be significantly altered or supported by AI during that period. While this might be reassuring for workers fearing automation, it presents a problem for companies pouring billions into AI technologies with expectations of dramatic productivity gains.
“Billions of dollars are at risk of being wasted,” Acemoglu says. “There will not be an economic revolution from that mere 5%.”
Acemoglu is one of the leading voices cautioning that the AI-driven stock boom and corporate spending spree may have gone too far. His academic career, which includes co-authoring the bestseller Why Nations Fail, has long focused on the economics of innovation and technology. While AI proponents suggest that businesses will see dramatic efficiencies and new breakthroughs, Acemoglu remains skeptical that the returns will be as large or as widespread as expected.
For example, Nvidia’s CEO, Jensen Huang, has stated that the demand for AI services will drive up to $1 trillion in data center investments. But despite this optimism, Acemoglu sees three possible future scenarios for AI’s role in the economy, none of which deliver the revolution many expect.
The first and most optimistic scenario is a “modest” rollout of AI, where its use grows incrementally and the excitement gradually cools. The second scenario sees the current hype peaking before a major tech stock crash, leaving investors and executives disappointed by AI’s slower-than-expected progress—an “AI spring followed by an AI winter.” The third, and most troubling, scenario envisions corporations continuing to chase AI at full speed, cutting jobs and funneling vast sums of capital into the technology without fully understanding its limitations. When AI’s limitations become clear, companies could be left scrambling to rehire workers and address economic imbalances, leading to broader negative consequences for the global economy.
Of these scenarios, Acemoglu leans towards a mix of the second and third as the most likely outcome. “There’s too much fear of missing out among executives to see the hype slowing down anytime soon,” he says, warning that “when the hype builds like this, the fall is unlikely to be gentle.”
Recent data supports his cautionary outlook. In the second quarter alone, tech giants such as Microsoft, Alphabet, Amazon, and Meta invested over $50 billion in AI-related capital spending. Despite these heavy investments, skepticism is growing as revenues haven’t matched costs in companies like Microsoft and Amazon.
So why aren’t today’s AI technologies, like OpenAI’s ChatGPT, living up to the hype in terms of labor impact? Acemoglu explains that while these models are impressive, they struggle with reliability and lack the human judgment needed to handle most tasks effectively. Tasks requiring complex decision-making or nuanced understanding, especially in white-collar roles, are unlikely to be outsourced to AI anytime soon. Additionally, AI remains unsuitable for automating physical labor-intensive jobs like construction or cleaning.
“You need highly reliable systems that can replicate the precise steps workers take,” Acemoglu notes. “AI can achieve this in a few limited areas with human oversight—such as coding—but in most places, it’s not yet a viable substitute.”
