U.S. labor productivity is growing at its fastest pace in decades, but it is still difficult to determine what role artificial intelligence has played. Much of the improvement may be linked to increased investment in equipment, more intensive use of resources, and business expansion that began after the pandemic.

This is an important question for companies investing hundreds of billions of dollars in AI in anticipation of substantial gains in economic efficiency. For now, there is insufficient compelling evidence that these investments have already delivered widespread results beyond the technology sector, Axios reports.

Productivity growth is not necessarily linked to AI

According to estimates from the Federal Reserve Bank of San Francisco, increased investment and more intensive use of labor and equipment have contributed 1.3 percentage points to annual productivity growth since the beginning of 2023. That is nearly twice their contribution over the previous two decades.

Excluding the effects of additional investment and more intensive use of resources, productivity grew by an average of 0.7% per year. By comparison, that figure was 0.5% in 2005–2022.

Some of the investment may indeed be linked to AI—for example, software that allows employees to complete more tasks in less time. But separating the impact of such technologies from other changes in how companies operate is not easy.

In the Federal Reserve's Beige Book report, businesses in the San Francisco district attributed productivity gains in roughly equal measure to AI and traditional improvements, including employee training and workflow optimization.

Companies still cannot precisely measure AI's contribution

Even organizations actively adopting new technologies acknowledge that it is difficult to separate their impact from the results of other changes.

Keith Waddell, head of staffing firm Robert Half, told investors that there is a significant gap between productivity gains for individual employees and efficiency gains across an entire organization. He noted, however, that the company is already seeing some benefits from using AI.

S&P Global expects to improve productivity by about 2% annually, but it also cannot determine precisely how much of that result is attributable to artificial intelligence.

“Some of this is directly related to AI, some is not. But most of it is the combined effect of integrating various changes,” the company's chief financial officer, Eric Aboaf, explained during its quarterly earnings call.

In other words, even if an employee completes individual tasks faster thanks to AI, this does not necessarily translate into a comparable increase in profits or revenue for the company as a whole.

The impact is more apparent at technology companies

For now, more obvious results are being seen at AI developers themselves. According to documents filed by Anthropic in preparation for its IPO, the Claude model wrote more than 90% of the code added to the company's codebase last month.

The average daily volume of code per engineer was more than 45 times its 2024 level.

However, such figures cannot automatically be extrapolated to other industries. Software development is one of the areas where generative AI has particularly strong capabilities, and an increase in the volume of code produced does not, in itself, mean a proportional improvement in product quality or economic returns.

The key test lies beyond the technology sector

For now, the financial performance of the largest technology companies looks more convincing than that of the rest of the U.S. stock market.

According to Apollo chief economist Torsten Slok, revenue per employee and profit margins have increased at the so-called “Magnificent Seven” largest technology companies, but have generally remained virtually unchanged at the other 493 companies in the S&P 500 index.

Slok believes that the results of these 493 companies, rather than those of the largest technology giants, will ultimately determine how successful AI investments prove to be.

If the technology truly can improve productivity across the economy, its impact should eventually show up in the financial results of companies across a wide range of industries. For now, however, the acceleration in the U.S. economy cannot confidently be regarded as proof that massive investments in artificial intelligence are already paying off.