The promise of artificial intelligence to revolutionize the workplace and drive economic growth is facing a familiar challenge: a disconnect between investment and measurable results. A new study by the National Bureau of Economic Research (NBER) reveals that despite widespread adoption of AI technologies, the vast majority of executives across the U.S., UK, Germany, and Australia report little to no impact on their operations.
The findings echo a phenomenon observed decades ago with the rise of computer technology, a period economists now refer to as “Solow’s productivity paradox.” In 1987, Nobel laureate Robert Solow famously observed, “You can see the computer age everywhere but in the productivity statistics.” The current situation, with AI increasingly visible yet failing to translate into significant productivity gains, has prompted a resurgence of that comparison.
The NBER study, which surveyed 6,000 CEOs, CFOs, and other executives, found that approximately two-thirds of firms are utilizing AI, but the average usage amounts to just 1.5 hours per week. A significant 25% of respondents reported not using AI at all. Critically, nearly 90% of firms surveyed indicated that AI had no discernible impact on either employment levels or productivity over the past three years.
Despite the current lack of measurable impact, executive expectations remain high. The study found that firms anticipate AI will boost productivity by 1.4% and output by 0.8% over the next three years. Interestingly, there’s a divergence in expectations between executives and employees, with executives forecasting a 0.7% decrease in employment while employees anticipate a 0.5% increase.
The disconnect between investment and outcome isn’t lost on economists. Apollo Global Management’s chief economist, Torsten Slok, recently wrote that “AI is everywhere except in the incoming macroeconomic data,” directly invoking Solow’s observation. Slok further noted a lack of evidence of AI’s impact in employment data, productivity figures, or inflation rates, and that, outside of the “Magnificent 7” tech companies, there are “no signs of AI in profit margins or earnings expectations.”
Academic research offers a mixed picture. A Federal Reserve Bank of St. Louis report from November 2023 suggested a 1.9% increase in cumulative productivity growth since the introduction of ChatGPT in late 2022. However, a 2024 MIT study, authored by Nobel laureate Daron Acemoglu, found a more modest projected increase of 0.5% in productivity over the next decade. Acemoglu acknowledged the figure was “disappointing relative to the promises” made by the technology industry.
Adding to the complexity, a 2026 Global Talent Barometer from ManpowerGroup revealed a 13% increase in workers’ regular AI apply in 2025, but simultaneously a concerning 18% decline in confidence in the technology’s utility. This suggests a growing skepticism among those directly interacting with AI tools.
IBM’s chief human resources officer, Nickle LaMoreaux, recently announced plans to triple the tech giant’s hiring of young employees, a move that suggests a recognition that while AI can automate certain tasks, a reliance on automation could create a shortage of middle-management talent and jeopardize the company’s long-term leadership pipeline.
The historical parallel to the IT boom of the 1970s and 80s, which ultimately led to a surge in productivity in the 1990s and early 2000s, offers a potential path forward. Slok suggests that AI’s impact may follow a “J-curve,” characterized by an initial period of slow gains followed by exponential growth. However, he emphasizes that this outcome hinges on the value created by AI and how effectively it is implemented across various sectors of the economy.
Unlike the IT era, where innovators enjoyed monopoly pricing power, the current AI landscape is marked by “fierce competition” among large language model developers, driving down prices. Slok argues that the future of AI productivity will therefore depend on companies’ willingness to embrace the technology and integrate it into their operations. “In other words, from a macro perspective, the value creation is not the product,” Slok said, “but how generative AI is used and implemented in different sectors in the economy.”