Almost everyone agrees that artificial intelligence has the potential to transform the economy in the coming decades. But no one is sure what impact the technology is currently having.
By some measures, AI is contributing to high unemployment rates among new college graduates and may have already eliminated tens of thousands of jobs. Other sources suggest that companies could actually hire new workers because of the technology.
AI could contribute to or be part of the solution to the US inflation problem. It could be responsible for the recent surge in productivity growth, or it could play virtually no role – or the productivity boom itself could be a mirage.
Researchers can’t even agree on basic questions, like how many companies are using AI or which workers are most vulnerable to the disruption it could cause.
The conflicting signals partly reflect the challenge of detecting economic changes in real time. Government statistics are retrospective in nature and can measure general trends better than developments in specific sectors or regions. New technologies that could lead to the creation of new products, jobs or entire industries can be particularly difficult to measure.
What sets AI apart from others is the speed at which it spreads throughout the economy. It took less than four years for generative AI to evolve from a novelty, mostly useful for writing limericks, to a powerful tool used by the world’s largest companies. Economists believe the technology will have profound effects on workers and the economy, although they disagree about what that impact will be. If the data is clear, they warn, it may be too late for policymakers to figure out how to respond.
“The stakes are very high,” said Nathan Goldschlag, research director at the Economic Innovation Group, a think tank. “The right policy depends on the right measurement. You can’t get the policy right if you don’t know what’s happening.”
Mr. Goldschlag released a report on Thursday documenting the challenge of AI measurement and suggesting steps to improve it. He and other experts argue that the government and private sector should devote more resources to the problem.
At least they get a hearing in Washington. In June, a bipartisan group of senators introduced a bill that would expand data collection and require the federal government to produce an annual report on the impact of AI on the workforce.
“The government has to make some important decisions about AI and the economy, and if you do that in a vacuum, you’re going to make mistakes,” said Sen. Mark Kelly, an Arizona Democrat and one of the bill’s sponsors. “This impacts the lives of millions of Americans and millions of businesses. And you can’t do it intelligently without reliable data.”
Table of Contents
Mixed signals
Policymakers are not completely blind. Since 2023, the Census Bureau has been surveying companies about their use of AI in a biweekly survey. Questions about technology were also included in an annual business survey, although only intermittently.
Researchers have developed several measures of “AI exposure,” many of which use a government database of job descriptions to assess which occupations will be most affected. Economists can use these measures to find out whether, for example, the most vulnerable occupations are slower to create new jobs or experience different rates of wage growth.
The problem is that the sources often tell confusing or contradictory stories. Surveys come to very different estimates of companies’ use of AI depending on how questions are asked. AI exposure measurements tell different stories about which jobs will be most affected. In a study, economists from Northwestern University and American University found that using different measures of exposure could influence not only the magnitude of AI’s impact on jobs, but also its direction. By some measures AI hurt employment, but by others it helped.
“It’s like going to the doctor and getting three different diagnoses for the same condition,” said Michelle Yin, an economist at Northwestern University and one of the study’s authors.
Part of the problem is that the most familiar metrics in economics were developed for a time before personal computers and the Internet, let alone AI. The Bureau of Labor Statistics’ monthly employment report, for example, provides breakdowns of job growth in manufacturing, retail and construction, but not in technology, where AI tools have been deployed most aggressively. Instead, technology is spread across multiple categories, including information, a broad sector that also includes newspapers and movie studios.
The jobs report contains even less information on jobs that may be at risk of displacement, such as software developers, accountants and customer service representatives. The most recent breakdown of detailed careers is from May 2025, a lifetime ago in the rapidly evolving world of AI
Still, economists say that despite its flaws, government data will be crucial to understanding the impact of AI over time. Researchers at the Yale Budget Lab, for example, have begun publishing a monthly analysis based on government data that tracks “job churn,” that is, how quickly the types of jobs in a given industry are changing. The measure is intended to be a kind of early warning system for the effects of AI. As companies begin adopting the technology, the theory goes, they will likely start filling other positions, even if their total number of employees doesn’t change immediately.
“It’s easy to pick up case studies after the fact,” said Martha Gimbel, the lab’s executive director. “What’s different this time is that we’re actually trying to measure this and figure this out in real time.”
But those efforts could be hampered by a federal statistical system that suffers from declining response rates to government surveys. Shrinking budgets are making it difficult for statistical offices to close the gaps. Erika McEntarfer, who led the Bureau of Labor Statistics until President Trump fired her last year, said $10 million a year in additional funding would allow the agency to expand the sample size of its monthly labor market survey so it can better capture economic changes.
“The data we currently use to understand the impact of AI on the labor market is at risk due to funding constraints,” she said. “It would only take some very modest investments to support them.”
Private data
Many economists aren’t waiting for the government to catch up. Several research teams have published AI measures based on private sector data that are more detailed and timely, although less comprehensive, than data available from the government.
Stanford University’s Digital Economy Lab released a dashboard of AI indicators last month based in part on data from payroll processor ADP. This data shows that entry-level jobs in the most AI-exposed sectors have declined sharply since ChatGPT launched in 2022. Erik Brynjolfsson, the lab’s director, called the trend a canary in the coal mine for AI-related job losses.
“I think it is comparable to the Industrial Revolution in terms of the impact on the labor market,” Brynjolfsson said. “I wish the federal government would invest more in it. But now there are some great private data sources that we’re compiling, and I think that’s what’s helping to close that gap.”
But the private data is just as unclear as the government statistics. A study released this week by Ramp, an expense management company, and Revelio Labs, a labor market data company, found that the companies using AI most intensively added new jobs faster than those that were slower to adopt the tools.
Ramp has access to data about which AI tools its customers buy and how much they spend on them. This makes it possible to distinguish heavy users from more cautious users – a crucial distinction because companies need time and investment to figure out how to use the tools effectively, said Ara Kharazian, senior economist at Ramp.
“It is difficult to measure the impact of AI on a business because it requires sustained adoption,” he said. “Our work makes it clear that a simple chat subscription does not increase a company’s productivity.”
However, such data is not necessarily representative of the entire economy. ADP’s customers tend to be relatively large and well-established. Ramp’s customers are generally tech-savvy. But if AI is to have the impact that its biggest supporters promise, it must be adopted by companies of all shapes and sizes.
In progress
Researchers generally agree on one thing: the impact of AI on the overall economy has so far been limited.
This isn’t necessarily surprising. Mr. Brynjolfsson and other economists have found that technological innovation often follows a J-shaped pattern, in which companies initially become less productive as they experiment with new tools and then make rapid progress once they figure out how to use them.
The confusing economic data suggests that many companies are still in downward J territory.
“The signals are mixed because probably the underlying economic conditions are mixed because we are still in a period of experimentation,” said Mr. Goldschlag, an economist at the Economic Innovation Group. “The tools themselves are still becoming useful.”
If there is indeed a mass disappearance of white-collar jobs, as some in Silicon Valley are predicting, it won’t take long for the losses to show up in government data. But even then, it may not be obvious that the AI is to blame.
The US economy has experienced a number of shocks unrelated to AI in recent years: the Covid-19 pandemic and its impact, including the struggles to return to office that continue to this day; inflation and the high interest rates the Federal Reserve has implemented to combat it; and drastic changes in government policy on immigration, trade and other areas. If a company has cut jobs since 2022, it’s not easy to tell whether that’s due to AI, high interest rates, or both.
Over time, it should become easier for researchers to separate the effects of AI from other forces. But they still won’t be able to answer the question that policymakers and ordinary citizens most want to answer: What’s next?
https://www.nytimes.com/2026/07/02/business/economy/ai-economy-data.html
