Congress has held hearings on AI. Tech CEOs have gone on tour warning that AI will remake every job in America within a few years. Meanwhile, a smaller group of economists at the Federal Reserve Bank of St. Louis has been doing something less dramatic. They asked actual workers and business owners what was happening on the ground.

The gap between the two stories is bigger than you'd think.

How Many Workers Are Using AI on the Job Right Now

Researchers Alexander Bick and Adam Blandin, working with co-authors from Vanderbilt and Harvard, ran what they describe as the first nationally representative survey of generative AI use at work. Their finding, covered in detail by The New Republic, surprised even the people running the study. AI use showed up in three out of every four occupations they measured, with at least a fifth of workers using it in each one. That's a wider spread than the researchers expected going in.

So adoption is real and it's broad. Somewhere between 43% and 45% of workers report using AI on the job to some degree, based on estimates discussed on a Richmond Fed podcast covering the research.

But here's the part that gets left out of most headlines about the AI jobs revolution. Almost nobody has handed an entire task over to AI. Workers who use these tools estimate their work moves along about 5% faster because of it. That's a real number worth having. It is also a modest one, especially next to the "this changes everything" framing coming from parts of the tech industry.

What Business Survey Data Reveals About AI Adoption Rates

A separate St. Louis Fed survey looked at businesses across its Eighth District, a region covering Arkansas and parts of six other states. The full write-up from the bank's economists, authored by Violeta Gutkowski, Charles Gascon, and Rehann Silvanus, breaks the numbers down clearly.

Only 34% of firms said AI tools are used regularly, and even then only by a small share of employees. Another 25% said they were still testing tools without regular use. And 11% reported no adoption at all. Put those together and roughly seven in ten businesses in the survey are either dabbling or standing still.

Firms that have gone further are seeing something worth paying attention to. One law firm told researchers that AI tools for legal research, deposition summaries, and first drafts saved each attorney two to three hours a week. A restaurant operator said AI order-taking let them skip hiring for a drive-thru position they otherwise would have needed. A professional services firm credited AI with supporting 35% revenue growth without adding staff.

That's genuine efficiency. It's the kind of gain a business owner should take seriously. It is not, however, evidence of mass job displacement.

Why Many Companies Still Haven't Adopted AI Tools

The survey also asked companies why they hadn't adopted AI, and the answers were pretty unglamorous. About 38% pointed to gaps in skills, data, or technical infrastructure. Another 34% said the tools on the market don't yet solve a real problem they have. Smaller numbers of respondents mentioned tight budgets, a workforce reluctant to learn new systems, or concerns about AI generating false information without enough oversight in place to catch it.

That skills gap is worth sitting with for a second. It's the single biggest reason companies give for standing still on AI, ahead of cost, ahead of trust issues, ahead of everything else.

This is exactly the kind of gap a vetted freelance marketplace like Begig is built to close. Instead of a company spending months hiring a full-time AI specialist it may not need past the implementation phase, it can bring in a freelancer who already knows how to set up the right workflows, train a team on the tools, and hand off something the in-house staff can maintain. That matches almost exactly what the survey found firms want: more output and capacity without a bigger permanent headcount.

What This Data Means for Staffing and Hiring Plans

Here's the number that should reassure anyone bracing for AI-driven layoffs, at least for now. Nearly half of all businesses surveyed, 49%, said they expect no noticeable change to staffing levels because of AI over the next 12 months. About 18% expect the skills they need to shift, without cutting headcount. Only around a fifth expect any staffing reduction tied to AI.

Compare that to what happened with the steam engine, the assembly line, or the personal computer. Those technologies eventually reorganized entire industries and job categories. AI hasn't done that yet, at least not according to the people actually running businesses and doing the work.

That could change. No one can promise it won't. But right now, the data points to a technology being adopted carefully and unevenly, not one sweeping through the economy overnight.

Cutting Through AI Hype With Careful, Sourced Data

A lot of the loudest AI predictions come from people with a stake in the outcome. Companies selling AI tools have an incentive to describe transformation just around the corner. Companies warning about AI risk are often the same ones racing to build it, which makes the warnings worth reading carefully rather than taking at face value.

So far, lawmakers have mostly held off on rushing through AI legislation, which lines up with what this survey data suggests. There's more time to get this right than the loudest voices imply. Regulation will probably be needed eventually. But grounding that conversation in what businesses and workers are actually reporting, rather than in speculation, is the better starting point.

If your team is one of the 70% still dabbling or standing still on AI adoption, that skills gap doesn't have to sit there unsolved. Browse vetted AI and automation freelancers on Begig and see what closing that gap actually looks like for a business your size.