Newfangled computer systems have been truly at occasions producing an excessive amount of info, producing agonizingly detailed studies and printing them on reams of paper. What had promised to be a growth to office productiveness was for a number of years a bust. This sudden consequence turned often known as Solow’s productiveness paradox, due to the economist’s remark of the phenomenon.
“You can see the computer age everywhere but in the productivity statistics,” Solow wrote in a New York Instances E book Evaluate article in 1987.
New knowledge on how C-suite executives are—or aren’t—utilizing AI reveals historical past is repeating itself, complicating the same guarantees economists and Massive Tech founders made in regards to the expertise’s impression on the office and economic system. Regardless of 374 firms within the S&P 500 mentioning AI in earnings calls—most of which mentioned the expertise’s implementation within the agency was fully constructive—based on a Monetary Instances evaluation from September 2024 to 2025, these constructive adoptions aren’t being mirrored in broader productiveness good points.
A examine revealed this month by the Nationwide Bureau of Financial Analysis discovered that amongst 6,000 CEOs, chief monetary officers, and different executives from corporations who responded to numerous enterprise outlook surveys within the U.S., U.Ok., Germany, and Australia, the overwhelming majority see little impression from AI on their operations. Whereas about two-thirds of executives reported utilizing AI, that utilization amounted to solely about 1.5 hours per week, and 25% of respondents reported not utilizing AI within the office in any respect. Practically 90% of corporations mentioned AI has had no impression on employment or productiveness during the last three years, the analysis famous.
Nonetheless, corporations’ expectations of AI’s office and financial impression remained substantial: Executives additionally forecast AI will enhance productiveness by 1.4% and enhance output by 0.8% over the subsequent three years. Whereas corporations anticipated a 0.7% reduce to employment over this time interval, particular person workers surveyed noticed a 0.5% enhance in employment.
Solow strikes again
In 2023, MIT researchers claimed AI implementation might enhance a employee’s efficiency by almost 40% in comparison with staff who didn’t use the expertise. However rising knowledge failing to point out these promised productiveness good points has led economists to marvel when—or if—AI will provide a return on company investments, which swelled to greater than $250 billion in 2024.
“AI is everywhere except in the incoming macroeconomic data,” Apollo chief economist Torsten Slok wrote in a latest weblog submit, invoking Solow’s remark from almost 40 years in the past. “Today, you don’t see AI in the employment data, productivity data, or inflation data.”
Slok added that exterior of the Magnificent Seven, there are “no signs of AI in profit margins or earnings expectations.”
Slok cited a slew of educational research on AI and productiveness, portray a contradictory image in regards to the utility of the expertise. Final November, the Federal Reserve Financial institution of St. Louis revealed in its State of Generative AI Adoption report that it noticed a 1.9% enhance in extra cumulative productiveness development for the reason that late-2022 introduction of ChatGPT. A 2024 MIT examine, nevertheless, discovered a extra modest 0.5% enhance in productiveness over the subsequent decade.
“I don’t think we should belittle 0.5% in 10 years. That’s better than zero,” examine writer and Nobel laureate Daron Acemoglu mentioned on the time. “But it’s just disappointing relative to the promises that people in the industry and in tech journalism are making.”
Different rising analysis can provide the reason why: Workforce options agency ManpowerGroup’s 2026 World Expertise Barometer discovered that throughout almost 14,000 staff in 19 international locations, staff’ common AI use elevated 13% in 2025, however confidence within the expertise’s utility plummeted 18%, indicating persistent mistrust.
Nickle LaMoreaux, IBM’s chief human assets officer, mentioned final week the tech large would triple its variety of younger hires, suggesting that regardless of AI’s means to automate a few of the required duties, displacing entry-level staff would create a dearth of center managers down the road, endangering the corporate’s management pipeline.
The way forward for AI productiveness
To make sure, this productiveness sample might reverse. The IT growth of the Seventies and ’80s ultimately gave technique to a surge of productiveness within the Nineteen Nineties and early 2000s, together with a 1.5% enhance in productiveness development from 1995 to 2005 following many years of hunch.
Economist and Stanford College’s Digital Financial system Lab director Erik Brynjolfsson famous in a Monetary Instances op-ed the development might already be reversing. He noticed that fourth-quarter GDP was monitoring up 3.7%, regardless of final week’s jobs report revising down job good points to only 181,000, suggesting a productiveness surge. His personal evaluation indicated a U.S. productiveness soar of two.7% final yr, which he attributed to a transition from AI funding to reaping the advantages of the expertise. Former Pimco CEO and economist Mohamed El-Erian additionally famous job development and GDP development persevering with to decouple consequently in a part of continued AI adoption, the same phenomenon that occurred within the Nineteen Nineties with workplace automation.
Slok equally noticed the longer term impression of AI as doubtlessly resembling a “J-curve” of an preliminary slowdown in efficiency and outcomes, adopted by an exponential surge. He mentioned whether or not AI’s productiveness good points would comply with this sample would rely on the worth created by AI.
Thus far, AI’s path has already diverged from its IT predecessor. Slok famous within the Nineteen Eighties, an innovator within the IT house had monopoly pricing energy till rivals might create related merchandise. Immediately, nevertheless, AI instruments are readily accessible because of “fierce competition” between massive language model-buildings driving down costs.
Due to this fact, Slok posited, the way forward for AI productiveness would rely on firms’ curiosity in making the most of the expertise and persevering with to include it into their workplaces. “In other words, from a macro perspective, the value creation is not the product,” Slok mentioned, “but how generative AI is used and implemented in different sectors in the economy.”
