Even before artificial intelligence threatened to destroy humanity, it was poised to wipe out the workforce. So said legions of doomsayers, from tech executives to computer scientists to Wall Street analysts, who gained notoriety predicting mass unemployment as thinking algorithms mastered the complex tasks once performed by workers.

Citrini Research, a small investment research firm, released a report in February warning of the swift arrival of a grim future, in which AI implementation leads to a double-digit unemployment rate. The viral report spooked investors: The S&P 500 fell more than 1% the day after its publication. More recently, Bill Gates issued a 6,000-word manifesto on the AI era, asking leaders to act quickly before "unemployment rises sharply."

Citrini's and Gates' AI stories imply that we are between a rock and a hard place. If the technology delivers all it promises, many white-collar workers will be redundant. If it fails, the economy and the stock market-both increasingly reliant on healthy returns from AI-related investments-could crumble. Either outcome could spoil the labor market.

But these dark prognostications fail to recognize that technological revolutions don't occur in vacuums. Factors like physical infrastructure, demographics, and the pace of implementation arguably matter more for the labor market and broader economic outcomes than any individual technological advance. Emerging data show that companies leveraging AI the most are implementing fewer layoffs, not more. And, encouragingly, new research suggests that AI is actually creating new roles.

The World Economic Forum projects that 92 million jobs will be displaced because of technological and economic shifts by 2030. But they also think there will be 170 million new ones, resulting in a net addition of 78 million jobs.

John Cochrane, a prominent economist and senior fellow at the Hoover Institution at Stanford University, says we should "stop worrying about the jobpocalypse."

"What happens when you get a cost-saving technology is that it expands output so much and new businesses expand so much that there is more jobs and everybody is better off," he says.

Cochrane notes that "everything since Gutenberg's printing press" has held to that trend. The internet is a prime example. In the late 1990s, job-related internet usage created productivity gains of roughly 0.25% a year. It ultimately created 2.6 jobs for every position it destroyed from 1994 to 2009, according to the McKinsey Global Institute.

"When you introduce technology and automate part of a worker's job, you might need fewer people doing that job," says Anu Madgavkar, a partner at MGI. "But you also tend to need more new things done as a consequence of bringing that technology."

It is too soon to say that AI will definitely chart that course, but early signs are promising.

So far, the negative impact of AI on the labor market has been minimal-it was responsible for just a 0.10 percentage point drag on aggregate unemployment last year, Goldman Sachs economist Elsie Peng estimated. And companies heavily investing in and deploying AI have actually grown their workforce by roughly 10% in the two years following adoption, according to Ramp Economics Lab's analysis of data from more than 70,000 U.S. businesses.

The field of radiology is currently undergoing this shift. Researchers at the Children's National Hospital in Washington, D.C., are pioneering the use of AI to help read magnetic resonance imaging scans from a new class of portable machines that can plug into a wall outlet. Those machines allow MRIs to be conducted more cheaply and in more settings, but it can't deliver crisp images. AI is needed to "see" the details within the scans that human eyes might miss.

This might spark fears that AI will eradicate radiologists, but data show the field isn't suffering job cuts. In fact, the opposite is true.

Since the introduction of ChatGPT in 2022, which many mark as the start of the AI revolution, employment at U.S. diagnostic imaging centers is up about 12%, according to data from the Bureau of Labor Statistics. That far outpaces total U.S. job growth and beats employment gains in the broader healthcare sector during the same period. Much of that growth is due to increasing demand as the population ages, but it shows AI is augmenting rather than replacing human work.

New technologies can also create new jobs by spurring entirely new businesses. During the early years of computers in the 1980s, roughly 14,000 new software firms were formed in the U.S. AI may now be doing something similar.

As of August, new business creation was up 25% compared with the end of 2024, according to data from the Census Bureau. Research from the Bank of America Institute finds that younger and lower-income entrepreneurs are driving much of that growth. Those entrepreneurs are also more often solo founders, possibly driven by technology advancements that make it easier to do more with less human capital.

"You can now make so much progress as one person and with relatively little in terms of resources," says Julian Weisser, founder of the Solo Founders Program, a San Francisco-based incubator for entrepreneurs.

Solo founders started 36% of new companies last year-up 53% from 2019 levels, according to data from Carta, a software platform for start-ups. Payment platform Stripe found that solo founders accounted for 63% of C corporations formed in the second quarter of 2026, marking a record high.

AI eliminates entry barriers by assuming tasks entrepreneurs would previously have needed others to complete. The workforce-management software firm Homebase recently found that small businesses using AI for scheduling and payroll tasks have saved an average 7.7 hours and $343 a month, or more than $4,100 annually.

"Businesses are becoming sort of supercharged with AI. Businesses that couldn't have afforded to bring on somebody full time can now afford to bring on AI to help," Weisser says.

One such business is FastLease, an online platform founded last year by 18-year-old solo entrepreneur Aden Nurie. FastLease markets commercial real estate buildings across the U.S.

"I built a lot of the product with AI tools to automate the work I was doing initially," Nurie says. "AI has made it tremendously easier to scale fast, particularly in comparison to my last company."

The pace of full-fledged AI adoption within existing businesses, however, has been less dramatic and far more gradual. As of May, only 6% of work hours are spent using generative AI, according to the Project on Workforce's GenAI Adoption Tracker, which involves 40,000 survey respondents. The data show that AI use led to an aggregate time savings of 2.2% among workers.

This slow adoption is in part because the physical infrastructure needed to support widespread adoption is still in development. Another speed bump is the challenge of changing long-held business practices across complex organizations. In a recent survey of business leaders in the finance industry by the tech firm Rogo, 64% said that internal change management is their biggest unresolved issue in the AI transformation.

For the companies that have figured out how to implement AI, resulting efficiencies haven't necessarily led to job cuts.

HelloFresh, a service that mails recipes and ingredients to doorsteps around the country, started implementing AI in 2024 to make it faster and easier to pack meal boxes. The company's compute power usage has grown so much that it had to boost its cooling capacity at its Phoenix packing facility earlier this year to keep temperatures at a food-safe 38 degrees.

Workers zoom around the freezer-like warehouse the length of five football fields filling cardboard boxes with ingredients. Behind the scenes, AI algorithms are calculating and assigning them optimal routes to grab those ingredients, sequencing order allocation, and tracking inventory levels.

"AI isn't here to take over the world," says Dan Seidel, HelloFresh's global chief operating officer. "It is here to optimize what we do."

The improved packing logistics have allowed HelloFresh to double the number of meal options for customers. That has contributed to a 13.7% year-over-year rise in order volume in the second quarter and allowed employee head count at HelloFresh's U.S. distribution centers to stay consistent, around 7,000.

"Everyone has this perception that robots and AI are going to take over the workforce, Seidel says. "But we are really leveraging AI and will continue to leverage AI to enhance the customer experience, not eliminate workforce."

Recent academic research finds that industries with higher AI adoption rates have experienced something similar to HelloFresh: the same workers, just with enhanced output.

AI gave businesses a boost in total productivity growth compared with rates from 2022 to 2025, according to draft research from the Brookings Institution. A Morgan Stanley analysis of public data found industries with high AI exposure contributed 1.7 percentage points of the 2.4% growth in output per employee in 2025. That is a significant increase from 2024, in which those industries contributed just 0.7 percentage-point to growth in worker output.

However, total factor productivity growth, which measures a broader view that includes both labor and capital and is considered a key factor in fundamental economic advancement, has remained modest over the past year. And at least some of the economywide productivity growth is attributable to factors other than AI adoption.

"Some of that boost may have been down to some of the innovations through the pandemic," says Brian Coulton, chief economist at Fitch Ratings. "At this stage, it is hard to say that [AI] is really transforming things."