American companies are pouring money into artificial intelligence, and the firms spending the most are growing their workforces the fastest. That runs directly against two years of predictions that AI would hollow out the office and wipe out entry-level jobs for young Americans. A new working paper is the first to measure what companies actually do with AI rather than guess at it, and the numbers turn the scare story on its head.
The Firms Betting Biggest on AI Are Hiring
The paper, “A New Look at AI’s Impact on Jobs,” comes from researchers at the corporate spending platform Ramp and the workforce analytics firm Revelio Labs. Across 21,559 U.S. firms, they found that companies adopting AI intensively grew headcount by about 10.2 percent over the two years after adoption. Companies that spent trivial amounts on AI showed no measurable change at all.
The authors sorted adopters by how much they spent per employee, and only the top tier, averaging around $33 per employee each month, saw the hiring gains. Firms dabbling at a couple of dollars a month looked no different from firms that never adopted. The growth tracks real investment, which tells a simple story. AI is helping the companies betting on it to expand, and expansion means hiring.
AI Is Opening Doors for the Next Generation
The entry-level result is the one that should greatly inform the ongoing debate. At the companies most heavily adopting AI, entry-level headcount grew 12 percent, and young workers became a larger share of the workforce. Every version of the AI panic centers on the same fear, that the technology will slam the door on young people before they can get a foot in it. The firms leaning hardest into AI are doing the reverse. They are hiring more of the next generation of workers and building the on-ramps that a healthy labor market and a growing economy depend on.
The gains are also broad and steady. They showed up six to twelve months after adoption, once firms had worked the tools into their operations, and they reached across engineering, sales, administration, finance, and customer service rather than landing in one corner of the org chart. That is the signature of AI working as a force multiplier. It makes workers more capable and lets the businesses employing them take on more, which is how a new technology is supposed to lift an economy and the people in it.
Why This Study Measures Something Real
Most of what the public has heard about AI and jobs rests on a workaround. Researchers usually cannot see which companies use AI, so they score occupations by how “exposed” they are to what a language model could theoretically do, then assume workers in those jobs are affected. The widely covered Stanford “Canaries in the Coal Mine” paper worked this way and concluded that workers aged 22 to 25 in exposed occupations had seen a 13 percent relative drop in employment.
This study takes a different route. Ramp can see the actual payments companies make to AI vendors, down to the dollar and the date, and Revelio reconstructs those same companies’ headcount from public professional profiles. That lets the authors watch the moment a firm starts seriously paying for AI and track what happens to its hiring afterward. An occupation’s theoretical exposure is a forecast. A company’s AI invoices are a record of what it actually did.
The Researchers Don’t Overclaim, and That’s the Point
The strength of this study is that its authors refuse to say more than their data supports. They are upfront that AI adopters are a selected group. These firms were already larger and faster-growing before they bought a single AI subscription. The authors built their entire method around the selection problem, comparing early adopters against firms that would adopt later rather than against the whole economy, precisely so the comparison holds up. That restraint is what separates this from the scarier headline studies, and it is why the findings are worth taking seriously. The paper has not yet been through peer review.
Panic Makes Bad Policy
Here is why this reaches beyond the seminar room. Fear is already being written into law. Proposals for AI-specific layoff notifications and algorithmic management mandates are advancing through statehouses on the premise that AI is a jobs wrecking ball, and the growing patchwork of state rules threatens to bury the firms doing the hiring in compliance costs. When the alarm runs ahead of the evidence, the innovation-killing red tape built on it punishes the very investment that is creating the jobs.
The honest reading of the data is that AI’s labor effects are gradual and uneven, still coming into focus, and that the American companies leaning in hardest are adding people. That is a reason for confidence in the free enterprise that has made America the world leader in AI, and a reason to be wary of regulating against a forecast. NetChoice has long argued that America’s lead depends on letting the technology grow without preemptive restrictions born of political theater, and this research is one more reason to hold that line.
The workers this debate claims to protect are best served by a labor market where American firms can invest and grow. Washington and the states can do right by American workers by keeping the country’s AI engine running, not stalling it over a fear the evidence does not support.
Image via Unsplash.