Is AI Coming for Your Job? The Fear Is Loud. The Data Is Quieter, and More Specific.

We build and run AI automation for a living. Every week we're inside the same tools that people are afraid of, watching what they can do and, just as often, what they can't. From there, the honest picture isn't "robots take everything" and it isn't "nothing to see here." It's narrower than either, and much more useful.

Here's where we stand, up front. AI is not wiping out work across the economy — the data says so plainly. But it is changing who gets hired, starting with the youngest workers in a specific set of jobs, and that deserves attention now, not after the fact. Companies are also using "AI" as a convenient label for layoffs that have other causes. Fear and hype are both getting in the way of seeing any of this clearly.

So let's do what we did last month. Set the rumors aside and look at what's actually measurable.

Part 1What the numbers actually show

Start with the big picture, because it's the part that gets left out. The Budget Lab at Yale has been tracking whether AI is reshaping the U.S. labor market since ChatGPT launched. Its conclusion is blunt: the mix of jobs Americans hold is not shifting in a way that lines up with AI. Measures of AI use show no connection to changes in employment or unemployment. People in the most AI-exposed jobs aren't staying unemployed any longer than they used to. As Yale's Martha Gimbel put it, "No matter which way you look at the data, at this exact moment, it just doesn't seem like there's major macroeconomic effects here."

Stanford's Digital Economy Lab, which studies payroll data covering millions of workers, agrees on the big picture: "no evidence of widespread, economy-wide job displacement." Keep that sentence in mind. It's the one fact almost no viral post will tell you.

Now the layoff headlines. They're real, too. According to Challenger, Gray & Christmas, which has tracked layoff announcements for decades, AI was the #1 reason companies gave for cutting jobs for five months running through July. So far this year, AI has been cited in about 112,700 announced cuts, roughly 24% of the total. That sounds apocalyptic until you read the rest of the same report. Total layoffs through July are down 41% from last year. Planned hiring is up 25%. In Andy Challenger's own words: "While AI is shifting the labor market, it is not dismantling it."

The Economist ran the numbers from the other direction on September 4th, and the result is the headline nobody is sharing: "The jobs apocalypse is postponed. An AI jobs boom is here." That same day, the Bureau of Labor Statistics reported that the economy added 162,000 jobs in August, far above expectations, with unemployment at 4.1%. The Economist puts AI-related layoffs at about 16,000 a month this year. That's real, but it's small next to the roughly 1.7 million workers U.S. employers let go in a typical month for every reason. Its bottom line: AI has so far created around 1 million new American jobs, against roughly 200,000 layoffs blamed on AI since mid-2023. That's about five jobs gained for every one lost.

So which is it, disruption or boom? Both. Fewer people are being laid off overall, more are being hired, and the new AI economy is adding jobs faster than it's cutting them. But when companies do cut, they increasingly say "AI." Hold that thought. It matters in Part 3.

Part 2Where the pressure is real

Here's the part the "nothing to see here" crowd gets wrong. There is one clear, measurable signal, and it's landing on the people with the least cushion: young workers just starting out.

The Stanford team calls it "canaries in the coal mine." In their August 2026 update, employment for 22- to 25-year-olds in AI-exposed occupations is 19% below where it would be if it had kept pace with their peers in less-exposed jobs. Experienced workers in those same jobs show no comparable gap. The gap was smaller when the team first flagged it in August 2025, and it has widened steadily since.

The details matter more than the headline number:

• It's hiring, not firing. The gap comes mostly from companies bringing in fewer young people, not from pushing people out. Nobody gets a layoff notice; the entry-level opening just never gets posted.
• It depends on how AI is used. The declines are concentrated in jobs where AI mainly substitutes for human tasks. Where AI mainly complements workers, employment is flat or rising, especially for experienced people.
• It isn't just a tech-industry story. The pattern holds even after removing tech companies and computer jobs, and after controlling for interest rates and remote work.

The researchers call these early warning signs, not proof of cause, and we'll respect that. But if you're 23 and trying to land a first job in customer service or junior coding, this isn't abstract. The first rung of the ladder is getting harder to reach.

It also helps to be precise about who is hurting. Young workers as a whole are holding up well. The Economist notes the gap between unemployment for 20- to 24-year-olds and the overall rate is near a multi-decade low. The damage is concentrated in specific jobs. Since January 2023, employment has fallen roughly 10% for customer-service workers and 15% for administrative assistants. That's a real hit to real people, but it's a targeted one, not a generational wipeout.

That's why the fear is climbing fastest among young adults. According to Pew, 55% of Americans under 30 now say they're more concerned than excited about AI, a majority for the first time. They aren't being irrational; the entry-level squeeze in exposed jobs is real. But most of them are bracing for a much bigger wave than the data shows so far.

Part 3What's overblown

Now the other side, because the hype does as much damage as the doom.

"AI" in a layoff announcement is not proof that AI took the jobs. Remember that 24% figure? Challenger itself warns the category is murky. As Andy Challenger put it: "Naming AI in a layoff announcement can win over investors while pushing current and prospective employees away." Saying "we're getting leaner with AI" plays better on an earnings call than "we over-hired in 2021" or "tariffs are squeezing our margins." Yale's Gimbel points to exactly those less flattering causes, including immigration shifts, tariffs, and policy uncertainty. Economists have a name for this: "AI washing." Oxford Economics found AI accounted for only 4.5% of announced U.S. job cuts in the first 11 months of 2025. It suspects some firms are dressing up layoffs "as a good news story rather than bad news, such as past over-hiring." Glassdoor's chief economist says the same: when a company says AI is why it's cutting, "that doesn't necessarily mean that's actually why."

"Could automate" is not "will replace." The scariest numbers you'll see are almost always capability estimates. MIT found current AI can technically handle tasks tied to nearly 12% of the workforce. Goldman Sachs estimated 6–7% of U.S. workers could be displaced if AI is widely adopted. Those are ceilings on what's technically possible, not forecasts of pink slips. Whether a task gets automated depends on cost, reliability, liability, and whether customers put up with it.

That last part is where the hype runs into reality. Klarna, the payments company, became the poster child for replacing people with AI. It said its assistant was doing the work of 700 customer-service agents. Then its own CEO admitted the all-in push toward AI support "resulted in lower quality work," and the company started recruiting human agents again. We see the same thing up close in our own builds. AI is very good at the first 80% of a task. The last 20% — judgment, context, the angry customer, the exception nobody planned for — is still where people earn their keep.

Part 4What actually happens to work

Here's the twist that almost nobody mentions. The most important line in the Stanford research isn't the 19%. It's the split underneath it. Where AI replaces the task, hiring falls. Where AI helps the person doing the task, employment holds or grows. The same technology produces opposite outcomes depending on how a business decides to use it.

That's not new. It's how every major technology has worked. MIT economist David Autor and his colleagues traced 80 years of job titles and found that most people working today are in job specialties that didn't exist in 1940. The new work came mainly from technologies that boosted what workers could produce. Technologies that simply automated tasks slowed new work down. Both forces are real, and they run at the same time.

And the new work is already showing up, just not where the doom headlines are looking. According to The Economist, the occupations closest to the AI boom — engineers, software developers, mathematicians, data scientists — have added roughly 730,000 jobs above trend since 2022. The Burning Glass Institute estimates about 1% of professional jobs are now "AI jobs," roughly a million positions. And remember last month's data centers? Construction spending on them rose 60% in a single year, pulling in electricians, HVAC techs, grid engineers, and machine technicians. Indeed finds data-center installation and maintenance jobs advertise wages about 40% higher than comparable work elsewhere. Some of the best new AI jobs don't involve a keyboard at all.

So the useful question isn't "will AI take jobs?" It's "which tasks inside a job will AI take, and what does the person do with the time that frees up?" A job is a bundle of tasks. AI rarely swallows the whole bundle. It takes a slice, usually the repetitive slice, and the job reshapes around what's left. Whether that reshaping means a better job or a thinner one is mostly a management decision, not a law of physics.

That's the part that should worry us, and it's also the part that should give us hope. The outcome isn't locked in by the technology. It's decided by people: business owners, managers, and workers themselves.

Part 5Where we land, and what to do about it

If you're worried about your own job:

• Learn the tools in your own field, now. The Stanford data says experienced people who use AI as a complement are holding steady. The safest place to be is the person who knows how to get good work out of these tools, not the person competing with them.
• Lean into the last 20%. Judgment, relationships, accountability, and handling the exception are the parts AI does worst. Make those the center of how you describe your value.
• Watch for the "AI" label on news about your employer. Ask what's actually changing in the work. Sometimes it's automation. Sometimes it's a cost cut with a better press release.

If you run a business, including a small one:

• Use AI to make your people better before you use it to replace them. It's the humane choice, and the data suggests it's also the smart one. Complement-style use is where employment and output grow together. And as Klarna learned, cutting the humans first can cost you quality you then have to buy back.
• Don't pull up the ladder. If AI handles the entry-level grunt work, rethink the entry-level job instead of eliminating it. The people who'd have learned the business in those roles are your future senior staff. A company that stops hiring 23-year-olds today will have no experienced 35-year-olds in 2038.
• Be honest about why you're cutting. Blaming AI for a layoff it didn't cause feeds exactly the fear that makes your remaining people less willing to adopt the tools.

---