Will AI Take My Job? What the Data Actually Says in 2026

My barber asked me last month, mid-fade, "D, is this AI thing coming for my job?"

I told him no. Nobody is building a robot that can line up a temple the way he does, and nobody wants one. Then he asked about his daughter. She's 23, she just finished a marketing degree, and she has sent out ninety-some applications since spring.

That one I couldn't answer as fast. Because the honest answer is: AI probably isn't taking her job. It might be taking the job she would have gotten first.

So let's do this properly. Here's what the best data says right now, where it disagrees, and what you can actually do about it.

The short answer: no mass layoffs, but a real squeeze at the bottom of the ladder

Two of the most careful research teams in the country track this every month, and as of their latest updates they agree on the big picture.

Yale's Budget Lab compares the whole U.S. job market before and after ChatGPT. Its October 2026 update still finds no clear sign that AI is reshaping who works where. The mix of jobs is changing at about the same pace it always has (Yale Budget Lab).

Stanford's Digital Economy Lab uses payroll records from ADP covering millions of workers. Its August 2026 update says the same thing about the overall economy: "no evidence of widespread, economy-wide job displacement" (Stanford Digital Economy Lab).

Then it found the part that should get everyone's attention.

The canaries: young workers in AI-exposed jobs

Stanford looked at workers ages 22 to 25 in the jobs most exposed to AI, like software development and customer service. Their employment is now 19% below where it would be if it had kept pace with young workers in less-exposed jobs. A year earlier that gap was 15%, and it has widened steadily since (Stanford, August 2026 paper).

Three details matter more than the headline number:

1. It's happening through hiring, not firing. People already in those jobs aren't being pushed out. New people aren't being let in.
2. Experienced workers in the same jobs show no comparable gap. Same occupation, same exposure to AI, different age, different result.
3. How AI gets used decides who loses. The declines are concentrated where AI substitutes for the human doing the task. Where AI complements the worker, employment is flat or rising, especially for experienced people.

Stanford's monthly tracker, last updated October 21, adds a useful contrast. Early-career software developers and customer service workers show substantial declines. Home health aides, a job AI barely touches, show employment increases for the youngest workers (Stanford Canaries Dashboard).

The researchers explain it this way. Young workers bring mostly codified knowledge: the stuff from textbooks and training manuals, which AI can now reproduce. Experienced workers carry tacit knowledge: judgment, relationships, knowing which customer is about to walk. AI can't download that yet (Stanford Accelerator for Learning).

The authors are careful to call these early warning signs, not proof that AI caused all of it. Pandemic over-hiring in tech and higher interest rates are part of the story too, and they tested for both. The pattern held.

So which jobs are shrinking and which are growing?

The World Economic Forum surveyed more than 1,000 employers who together employ over 14 million workers. Their outlook for 2025 to 2030, worldwide:

• 170 million jobs created and 92 million displaced, for a net gain of 78 million (World Economic Forum).
• AI and information processing alone are expected to create 11 million jobs and displace 9 million (WEF Future of Jobs Report 2025).
• The roles employers expect to shrink fastest include cashiers, ticket clerks, administrative assistants, printing workers, and accountants and auditors (WEF). Graphic designers and legal secretaries made that fast-decline list for the first time ever (WEF Future of Jobs Report 2025).

Remember that's a forecast from employers, made in late 2024. It's what bosses planned, not what happened. But it lines up with what Stanford is seeing in real payroll data: the work most at risk is routine, rule-following, screen-based work, and especially the beginner version of it.

What it looks like here in Washington

Washington is a special case, because so much of our economy is tech.

• Jobs: Washington's unemployment rate fell to 4.8% in September 2026, its third straight monthly drop. But the state still lost an estimated 15,400 jobs that month, and the slowdown has hit hardest in the information and professional services sectors, where much of the Seattle tech workforce sits (MyNorthwest, citing the Employment Security Department).
• Tech layoffs: Amazon, Microsoft, Oracle and other tech companies in the state cut more than 30,000 jobs in 2025, with more in 2026. The state's own STEM report says it's still unclear how much of that is about AI (Washington STEM Education Report Card 2026).
• Demand that isn't going anywhere: that same report counts nearly 17,800 expected job openings a year in computer and IT jobs, against about 6,300 certificates and degrees completed in 2023–24. It notes those projections were built on trends from before the latest AI advances (Washington STEM Education Report Card 2026). Health care remains one of the state's steadiest sectors, and registered nurses are still among the most in-demand jobs (MyNorthwest).

When KIRO asked the state's chief labor economist, Anneliese Vance-Sherman, which skill would protect workers over the next five years, she wouldn't name one, "except curiosity and a commitment to lifelong learning" (MyNorthwest). I've heard a lot of answers to that question. That's the most honest one.

What Washington law does (and doesn't) do for you

Washington's AI Task Force released its final report on July 1, 2026, with 11 recommendations, plus a survey of Washington workers on AI at work (Washington Attorney General). Lawmakers passed four of those recommendations in full or in part. A bill regulating AI in high-stakes decisions like hiring died without a floor vote, and so did the recommendation to write guidelines for AI in the workplace (GeekWire).

So right now, if an algorithm screens your résumé or flags you at work, Washington doesn't give you a specific right to know or appeal. California and Connecticut have moved further; I broke that down in Your Boss's Algorithm Now Needs a Human.

The 20-minute task test

Forget "will AI take my job." Jobs are bundles of tasks, and AI takes tasks. So look at yours.

Step 1. List everything you did last week. Not your job title. The actual tasks. "Answered 40 customer emails." "Built the Monday report." "Calmed down a client who was ready to cancel." Aim for 15 to 25 items.

Step 2. Mark each one with a letter.

• R for rules: it follows a clear pattern, starts and ends on a screen, and someone could write the steps down. Drafting routine emails, summarizing documents, data entry, first-draft code, basic scheduling.
• J for judgment: it depends on context you carry in your head. Knowing which vendor always ships late. Reading the room in a meeting. Deciding what to leave out of a report.
• P for people: it needs trust, presence or hands. Coaching, caregiving, negotiating, teaching, fixing things, showing up.

Step 3. Do the math. If more than half your list is R, your role is exposed, and the beginner version of it is where hiring is drying up. That's not a reason to panic. It's a reason to move toward J and P on purpose.

Step 4. Make AI your assistant, not your replacement. Here's the most useful finding in the Stanford research: jobs where people use AI to help with their work are holding up. Jobs where AI does the work instead of the person are shrinking. So take one R task this week and hand it to an AI tool. Then spend the time you save on a J or P task. That's exactly the trade employers are paying for.

If you're just starting out

This is the part I'd tell my barber's daughter.

• Get the reps AI can't fake. Tacit knowledge comes from doing real work next to experienced people. Internships, apprenticeships, volunteer projects with a nonprofit, a part-time role at a small business. Any of these beats a 100th application.
• Show judgment, not just skills. A portfolio piece that says "here's the problem, here's what I decided and why" beats a list of tools.
• Learn the AI tools in your field, out loud. Put them on your résumé, and be ready to explain when you wouldn't trust them.
• Make sure the screener can read you. Much of entry-level hiring runs through automated systems first. I wrote a full guide: How AI Résumé Screeners Work and How to Get Seen.
• Consider the doors that are still wide open. Health care, skilled trades, and data jobs where you bring domain knowledge. If data interests you, start with Break Into Data Without a CS Degree.

If you're mid-career

The data is kinder to you, but don't coast.

• Write down what you know. The judgment in your head is your edge. Teach it, document it, mentor someone with it. People who make teams better are the last ones cut.
• Be the person who brings AI in safely. Every organization needs someone who can say "this tool is fine for drafts, but a human checks anything that goes to a customer." That person is hard to replace.
• Ask your employer three questions: What AI tools are we using on our own work? Will anyone lose hours or roles? What training comes with it?

The bottom line

AI isn't coming for everyone's job. It's coming for tasks, and right now it's hitting hardest the tasks we used to hand to beginners. That's a problem for young people trying to get in the door, and a challenge for every employer who'll need experienced people in ten years and isn't training any today.

The best move for all of us is the same: know which parts of your work are rules, judgment and people, and keep moving toward judgment and people.

Pour me another: Have you run the task test on your job? Tell me what you found through the contact form. The best answers might show up in the next Tea Report.

Sip slow, stay curious.D