AI Productivity Paradox Resurfaces as CEOs Admit Minimal Impact on Workforce

Across corporate boardrooms from New York to Frankfurt, a quiet revelation is reshaping the narrative around artificial intelligence: despite billions in investment and widespread adoption, AI has yet to deliver measurable gains in productivity or employment. According to a landmark study by the National Bureau of Economic Research (NBER), nearly 90% of the 6,000 C-suite executives surveyed across the U.S., U.K., Germany, and Australia reported that AI had no discernible impact on their firms’ operations over the past three years. The findings, first reported by Fortune and corroborated by Finance Yahoo, have reignited a decades-old economic puzzle known as Solow’s Productivity Paradox.

In 1987, Nobel laureate Robert Solow observed that while computers were everywhere, their impact on productivity statistics was conspicuously absent. "You can see the computer age everywhere but in the productivity statistics," he wrote. Today, economists are drawing direct parallels. Despite 374 S&P 500 companies touting AI in earnings calls as a transformative force, macroeconomic data tells a different story. Average AI usage among executives is just 1.5 hours per week, and a quarter of respondents admit to not using AI at all. Meanwhile, corporate spending on AI technologies surpassed $250 billion in 2024, according to Fortune.

The disconnect between expectation and reality is stark. While executives forecast a 1.4% productivity boost and a 0.7% reduction in employment over the next three years, workers themselves anticipate a 0.5% increase in job opportunities—a divergence that underscores a deeper uncertainty in how AI is being deployed. "AI is everywhere except in the incoming macroeconomic data," wrote Apollo chief economist Torsten Slok in a recent analysis. "Today, you don’t see AI in the employment data, productivity data, or inflation data."

Academic research offers conflicting signals. A 2023 MIT study claimed AI could boost individual worker performance by up to 40%, while a Federal Reserve Bank of St. Louis report noted a 1.9% cumulative productivity gain since ChatGPT’s 2022 debut. Yet, MIT economist and Nobel laureate Daron Acemoglu tempered enthusiasm, calling a projected 0.5% productivity increase over a decade "better than zero," but far below industry hype. "The promises made by tech journalists and venture capitalists have been wildly inflated," he said.

One emerging explanation lies in implementation. ManpowerGroup’s 2026 Global Talent Barometer found that while 13% more workers are using AI regularly, confidence in its utility dropped 18%. Employees report feeling overwhelmed by poorly integrated tools, inconsistent training, and a lack of strategic alignment. IBM’s Chief HR Officer Nickle LaMoreaux highlighted a structural risk: as AI automates entry-level tasks, companies risk depleting their pipeline of future middle managers. "We’re tripling our young hires because we need the human capital to manage the systems we’re building," she said.

Yet, hope persists. Stanford’s Erik Brynjolfsson points to early signs of a productivity surge, citing a 2.7% U.S. productivity jump in 2025—mirroring the delayed boom of the 1990s IT revolution. Slok suggests AI may follow a "J-curve": an initial plateau followed by exponential gains once companies master integration, workflow redesign, and workforce upskilling. "The value isn’t in the model," Slok argues. "It’s in how it’s used."

For now, the paradox endures. AI is not failing—it is being underutilized. The lesson from history is clear: technology alone does not drive productivity. Culture, training, and organizational change do. As the world waits for the next wave of AI-driven growth, one truth remains: the machines are ready. The human systems are not.