Picture two 23-year-old who graduate in the same year. Same city, same grades, same nervous energy outside the interview room.
One applies for a junior role in a field where AI tools already draft text, write code and answer customer chats. The other applies for a role built on hands-on judgement and practice. According to Stanford’s latest payroll research, young workers in highly AI-exposed occupations are now about 19% behind their peers in less-exposed ones, so the first applicant is walking into a much tighter door.
That gap is the real story behind every scary headline about Gen Z and AI. It is not a story about millions of young people fired by robots. It is a story about a career ladder where the bottom rung quietly vanished.
The two get mixed up all the time. Losing a job and never getting the first one feel the same on social media, but they are different problems with different fixes.
In this guide, I walk through 9 data points from Stanford, Gallup, the U.S. Census Bureau and UK research, in plain English. I also flag what the numbers cannot prove, because the researchers themselves are careful about that.
Table of Contents
- The Difference Between Losing a Job and Losing the First Rung
- Data Point 1: Young Workers in AI-Exposed Jobs Are 19% Behind
- Data Point 2: Researchers See No Widespread Job Displacement
- Data Point 3: The Squeeze Happens at the Hiring Door, Not the Exit
- Data Point 4: Jobs Where AI Automates Are Hit Hardest
- Data Point 5: Textbook Knowledge Is Losing to Experience
- Data Point 6: One in Three UK Employers Cut Entry-Level Roles
- Data Point 7: Over 1 Million Young Britons Are Outside Work and Study
- Data Point 8: Gen Z Uses AI Less Happily Than You Think
- Data Point 9: The Honest Caveat Nobody Puts in the Headline
- What Gen Z Can Do Right Now
- FAQ
- Final Thoughts
The Difference Between Losing a Job and Losing the First Rung
A layoff is loud. There is a notice, a headline and a crowd of people sharing the same story online.
A missing first rung is silent. Nobody announces it. A graduate sends 80 applications, hears nothing, and quietly assumes the problem is personal.
It is true that some big companies have announced large cuts and pointed at AI, and my post on Big Tech layoffs blamed on AI covers that wave in detail. But the broader data tells a different, quieter story about who is actually being squeezed.
Here is the simple difference:
| Mass layoffs | Lost first rung | |
|---|---|---|
| Who is affected | People already employed | People trying to enter the job market |
| How it shows up | Layoff notices, severance | Fewer openings, silence after applying |
| Media visibility | Very high | Very low |
| Main data signal | More workers leaving jobs | Fewer young workers being hired |
| Who feels it first | Existing staff | Graduates and early-career workers |
The first rung matters more than it looks. A junior job is where people learn how real work feels, how a good manager talks, and what a mistake costs. Remove it, and the whole ladder gets harder to climb for everyone who comes after.
Data Point 1: Young Workers in AI-Exposed Jobs Are 19% Behind
Start with the strongest number. The Stanford Digital Economy Lab tracks payroll data from ADP, one of the largest payroll processors in the United States. In its August 2026 update, the lab reports that workers aged 22 to 25 in highly AI-exposed occupations now stand about 19% below where they would be if they had kept pace with similar-aged workers in less-exposed jobs. Experienced workers show no comparable gap.
The gap has also grown. It was about 15% in the July 2025 data and reached about 19% by June 2026.
| Data vintage | Gap for ages 22 to 25 in highly AI-exposed jobs |
|---|---|
| July 2025 | about 15% |
| June 2026 | about 19% |
Here is what that looks like in raw terms. Between November 2022 and June 2026, employment of 22 to 25-year-olds in the two most exposed groups of occupations fell about 11%, while the same age group in the three least-exposed groups grew about 10%.
So it is not that young people stopped working. Young people in some fields kept finding jobs, and young people in others did not.
If you want a wider list of careers under pressure, my breakdown of the high-paying jobs AI is expected to erase first goes deeper. Just keep in mind that “under pressure” and “already gone” are not the same thing, which is exactly the point of this article.
Data Point 2: Researchers See No Widespread Job Displacement
This is the part most headlines skip. The same Stanford team says plainly that it does not see widespread, economy-wide job displacement linked to AI.
That fits the title of this post. The labor market overall is not collapsing. The pressure is concentrated in one age group and one type of work.
The U.S. government data points the same way, because a Census Bureau working paper shows broadly consistent raw patterns by age and industry, which the Stanford team cites as supporting evidence.
You can also follow the numbers as they change. Stanford and ADP now run a Canaries Dashboard that refreshes the key results every month, so you can watch the trend yourself.
So the fair summary is this: the economy is still hiring, but it is hiring fewer young people into some AI-exposed roles.
Data Point 3: The Squeeze Happens at the Hiring Door, Not the Exit
How is this gap created? Not mainly through firings. The Stanford team finds that the adjustment works mostly through reduced hiring of young workers, not through more people losing their jobs.
That single detail explains why the problem feels invisible. Nobody gets a dramatic goodbye email. Companies simply post fewer junior roles, or none at all, and the people who would have filled them never show up in a layoff statistic.
There is one more detail worth knowing. So far, the change shows up in whether young people get jobs, not in how much base pay they receive. In other words, the issue is getting through the door, not a pay cut once you are inside.
Executives have warned about this shift for a while, and my post on Big Tech CEO warnings about AI and jobs tracks those statements. Payroll data adds something warnings cannot: measurement.
Data Point 4: Jobs Where AI Automates Are Hit Hardest
Here is a detail that changes how you should read the whole story. Not every AI-exposed job is shrinking for young workers. What matters is how AI is being used inside that job.
Stanford’s researchers split AI use into two buckets. In the first, AI automates human tasks, such as writing code or handling customer chats. In the second, AI complements workers by helping them do the job better. The declines for young workers are concentrated in the first bucket, while in jobs where AI mostly complements people, employment is flat or rising, especially for experienced workers.
| How AI is used in the job | What happens to young workers | What happens to experienced workers |
|---|---|---|
| AI automates tasks | Employment falls | Little change |
| AI complements workers | Flat or rising | Flat or rising, often stronger |
That is a hopeful finding hiding inside a worrying one. The technology itself is not the villain in every case. The problem appears when a company uses AI to replace the small, repeatable tasks that juniors used to do.
This also gives Gen Z a practical lesson. Using AI to skip the learning is one thing. Using AI to learn faster and take on bigger work is another.
If you want a wider look at how AI is slowing down and reshaping jobs and apps, my post on how AI is changing jobs and apps adds useful background. The key idea to carry forward is simple: the same tool can open a door or close one, depending on who is using it and why.
Data Point 5: Textbook Knowledge Is Losing to Experience
This is my favourite finding in the whole study, because it explains why the first rung is the part that breaks.
The researchers separate two kinds of knowledge. Codified knowledge is formal and written down: the things you learn from textbooks, lectures and documented procedures. Tacit knowledge is the kind you pick up by doing the work, watching people who are good at it, and seeing real situations again and again.
According to a report on the revised Stanford paper, occupations that rely heavily on codified knowledge tended to see slower employment growth for entry-level workers. Jobs that depend more on tacit knowledge, meanwhile, showed stronger growth for mid-career and senior workers.
Why would that happen? The Stanford team says this pattern is consistent with a world where generative AI is especially good at reproducing knowledge that has already been written down. Knowledge built through practice and mentorship is much harder for a model to copy.
Think about what that means for a fresh graduate. A degree mostly hands you codified knowledge. That is exactly the kind AI handles best, and that is why the entry-level slice of those jobs looks weaker.
There is one more note worth flagging. The researchers also found that women face greater AI exposure on average, and they plan to keep watching that gap as new data arrives.
None of this says education is worthless. It says education alone is no longer a complete answer, and that hands-on experience has become the scarce, valuable ingredient.
Data Point 6: One in Three UK Employers Cut Entry-Level Roles
Now let us leave the United States and look at the UK, where the pattern shows up in a very different kind of evidence.
A survey of 1,001 businesses by Survation for Lancaster University’s Work Foundation, reported by Reuters, found that 36% of firms had reduced the number of entry-level jobs over the past year. The survey ran from May 7 to May 20, 2026.
The gap by company size is large:
| Employer size | Cut entry-level jobs | Said AI and automation lowered job numbers |
|---|---|---|
| Small businesses | 24% | 24% |
| Medium businesses | 48% | 58% |
| Large businesses | 46% | 60% |
Also, 73% of employers said it is harder for young people to find work than it was five years ago. Almost half of larger employers reported fewer vacancies than a year earlier, which means hiring is slowing beyond just the entry level.
The Work Foundation’s director, Ben Harrison, described it as one of the toughest labour markets in years for young people.
There are two honest cautions here. First, this is a survey of what employers say, not a payroll count. Second, the data says AI contributed “in part,” so it should not be treated as the only cause.
Still, I want to point out a risk that I think gets too little attention. If big companies automate the trainee tier today, who becomes the experienced manager in five years? That “hollow middle” problem is my own reasoning rather than a measured result, but it is the kind of long-term cost that some forecasters worry about, and my post on AI 2027 predictions shows how quickly some experts think these shifts could arrive.
Data Point 7: Over 1 Million Young Britons Are Outside Work and Study
The next number comes from the same UK labour market, and it is the one that makes people stop scrolling.
In the first quarter of 2026, more than 1 million people aged 16 to 24 in the UK were counted as NEET, which means not in education, employment or training. That is the highest reading since late 2013. The same figure appears in this report on the UK entry-level survey, which also notes that a government report in May warned of a “lost generation” of young people.
Now the honest part. NEET numbers have many causes, including a weak economy, health problems, education choices and local job shortages. AI is one piece of that picture, not the whole puzzle.
So why include it? Because it shows what a shrinking first rung can look like at national scale. When employers cut entry-level roles and young people cannot find a way in, some of them end up outside the system entirely.
Here is how I would read the three UK numbers together:
| UK data point | What it tells us | What it cannot tell us |
|---|---|---|
| 36% of employers cut entry-level roles | Hiring at the bottom is shrinking | How much of it is caused by AI |
| 73% say it is harder for young people to find work | Employers feel the squeeze too | Which sectors are worst |
| 1 million+ NEET aged 16 to 24 | A large group is outside work and study | Whether AI is a major driver |
The pattern is worrying, but it is a pattern, not a verdict.
Data Point 8: Gen Z Uses AI Less Happily Than You Think
Here is where the story gets surprising. You might expect the generation that grew up online to be the loudest fan of AI. The data says something more mixed.
A 2026 Gallup survey for the Walton Family Foundation and GSV Ventures asked 1,572 people aged 14 to 29 about AI in late February and early March. Just over half use it daily or weekly, with 22% using it daily and 29% weekly. About 19% say they never use it.
Use is steady. Feelings are not. Excitement about AI sits at 22% and hopefulness at 18%, both already low and now lower than a year earlier, while anger rose to 31% and anxiety stayed high at 42%.
| Feeling about AI (Gen Z, 2026) | Share |
|---|---|
| Curiosity | 49% |
| Anxiety | 42% |
| Anger | 31% |
| Excitement | 22% |
| Hopefulness | 18% |
Even heavy users are cooling off. Gen Z people who use AI every day are 18 points less excited and 11 points less hopeful than they were last year.
The workplace numbers are just as telling. Gen Z workers place more trust in work done without AI (69%) than in AI-assisted work (28%). According to the full Gallup and Walton Family Foundation report, nearly half of employed Gen Z say the risks of AI at work outweigh the benefits.
Here is the paradox in plain words. Many young people dislike how AI is changing things, yet they feel they have to learn it to stay competitive. Fifty-six percent of high school students now say they will have the skills to use AI in daily life after graduation, up 12 points from last year.
One more encouraging sign: the share of K-12 students who say their school has AI rules jumped from 51% in 2025 to 74% in 2026. Schools are catching up.
There is also a trap hiding inside this data. If Gen Z leans on AI for every task, it may skip the very practice that builds the experience employers value most, a pattern I describe in my post on the signs you are addicted to AI content tools. Use it as a coach, not a crutch.
Data Point 9: The Honest Caveat Nobody Puts in the Headline
If a blog post gives you only one side of a number, be suspicious. So let me give you the strongest warning against my own title.
The Stanford researchers say clearly that their numbers are descriptive patterns, not proof that AI caused the change. They write that they cannot yet answer the causation question definitively.
They also listed reasons to be careful. The gap between more and less exposed young workers shrinks once you account for education. Some differences were visible even before generative AI became widespread. And the gaps are larger in the ADP sample than in national survey benchmarks, which raises questions about how well the results generalize.
That matters for how you share this story. “AI caused a 19% drop” would be wrong. “Young workers in AI-exposed jobs are 19% behind their peers, and AI may be one of the reasons” is the honest version.
The team also tested other explanations. The divergence persists after excluding tech firms and computer jobs, and after accounting for exposure to interest-rate increases and remote work. Stanford even published a separate note on interest rates and timing as part of this question.
So we are left with a fair conclusion. AI is a plausible and meaningful suspect, but it is not a proven culprit.
I think this is exactly how AI claims should be treated, whether they sound alarming or reassuring, and my article on why some superintelligence risk claims get debunked makes the same point from another angle. Good evidence beats a dramatic headline every time.
What Gen Z Can Do Right Now?
The data describes a squeeze, not a sentence. Here is a practical plan. This section is my own advice based on the findings, not something the studies measured.
1. Aim for the “AI complements” side of your field
Stanford found that employment is flat or rising where AI helps people do their work, and weaker where it automates tasks. So before choosing a role, ask one question: in this job, does AI replace the beginner’s tasks, or does it make the beginner more capable? The Occupational Outlook Handbook from the U.S. Bureau of Labor Statistics is a good place to compare what different careers actually involve day to day.
2. Collect tacit knowledge on purpose
Remember the codified vs tacit finding. Textbook knowledge is the easiest for AI to copy, so practice is your edge. Look for internships, apprenticeships, volunteer projects, job shadowing and small freelance gigs. Even a few months beside someone skilled teaches things no course can.
3. Show proof of work, not just a degree
A degree says you studied. A portfolio says you can do. Build two or three real projects, write up what you learned, and keep the results in one place a recruiter can open in 30 seconds.
4. Use AI as a coach, not a crutch
Try this rule: do your first draft yourself, then ask AI to critique it, find gaps and explain what you missed. That way the tool speeds up your learning instead of replacing it. It also answers the worry in the Gallup data, where many young people fear AI may weaken their thinking.
5. Do not ignore the skills pathways beyond college
Not every strong career starts with a university degree. Gallup’s Voices of Gen Z research hub includes a report on Gen Z’s limited awareness of non-college pathways, which is worth a read if you are weighing options.
6. Build people, not only profiles
When formal entry-level roles shrink, introductions matter more. A short, honest message to someone doing the job you want can open doors that an online application form never will.
For employers and managers
If you run a team, the hollow middle risk is worth a serious thought. Companies that automate every junior task today may find they have no trained people to promote later. Keeping even a small apprenticeship pipeline is an investment in your own future bench.
Frequently Asked Questions
Is AI taking away all entry-level jobs?
No. Stanford’s researchers do not see widespread, economy-wide job displacement from AI. What they see is weaker hiring of young workers in certain AI-exposed occupations.
How big is the gap for young workers in AI-exposed jobs?
In the August 2026 Stanford update, workers aged 22 to 25 in highly AI-exposed occupations were about 19% below where they would be if they had matched similar-aged workers in less-exposed jobs. That gap was about 15% a year earlier.
Did AI definitely cause this?
Not proven. The researchers describe their findings as descriptive patterns, and they say they cannot yet answer the causation question definitively. They do think AI may be playing a meaningful role.
Which jobs are safest for Gen Z?
There is no guaranteed safe job. The research suggests roles that rely on practice, mentorship and real-world judgement, and jobs where AI helps workers rather than replaces tasks, have fared better so far.
Should Gen Z stop using AI?
No. Gallup finds that many young people feel they need AI skills even while feeling anxious about it. The smarter move is to use AI to learn and build faster while still doing real, hands-on work.
Final Thoughts
Let us return to the two 23-year-olds from the start. Their paths split not because one worked harder, but because the door they chose was narrowing.
The title of this post is the cleanest summary I can give you. Gen Z is not being fired in huge numbers. The first rung is getting harder to reach, mostly through quieter hiring, and AI is a plausible but unproven part of the reason.
That means the right response is not panic. It is aiming at roles where AI amplifies you, building the practical experience machines cannot copy, and demanding honest numbers from every headline, including this one. You can keep tracking the trend through Stanford’s AI Economic Indicators, which updates its key measures regularly.
Read the data, stay curious, and keep climbing.
Related reading:
If you want to understand where this technology could be heading, start with my look at the Elon Musk superintelligence quotes and what they really mean.
You can also read the warnings about AI as humanity’s last invention to see why some experts are worried.
For the other side of the debate, my breakdown of superintelligence extinction risk explains the arguments in plain English.
And if you are following the model race, see what the GPT-6 Astra AGI claim actually showed.