0
0
You send the application. Nothing comes back.
Not a no. Not a maybe. Nothing. You open the sent folder a week later just to check that it actually went.
Do that thirty times and something shifts. It stops feeling like bad luck and starts feeling like a verdict — like the world looked at you closely and decided you are not needed yet.
Entry-level jobs really are getting harder to find, and there is solid data on it. But the data says something quite different from what that silence is telling you.
The AI Jobs Apocalypse Did Not Show Up in the Numbers
For three years the story has been simple and loud: the machines are taking the work.
Economists at Stanford went looking for it. In a policy brief published in July 2026, Neale Mahoney, Erika McEntarfer and Karsen Wahal compared workers in the jobs most exposed to artificial intelligence against workers in the jobs least exposed to it.
If AI were quietly emptying offices, the gap would be obvious.
It wasn’t. Since 2022, unemployment among the most AI-exposed workers rose by 0.77 percentage points. Among the least exposed, it rose by 0.85 — slightly more. The people supposedly safe from the robots did marginally worse than the people supposedly in their path.
Their conclusion was that this looks like a broadly softening labour market rather than one shaped by AI-driven job losses.
That is worth sitting with for a second. The thing almost everyone believes is happening is not visible in the aggregate numbers at all.
One note before going further: this is United States data. The specific figures belong to one country. The shape of what follows travels much further than that.
The Exception: Entry-Level Jobs for People Who Haven’t Started
The same brief names one real exception, and it is a narrow one.
Unemployment among recent graduates reached 5.6% in early 2026 — up 1.6 percentage points in three years. Not a catastrophe. But a clear, measurable move in one direction, concentrated on one group.
A separate Stanford team has been tracking this closely. The Digital Economy Lab’s August 2026 update found that employment among workers aged 22 to 25 in the most AI-exposed occupations now sits about 19% below where it would be if it had kept pace with similarly aged workers in less-exposed jobs. In July 2025 that gap was 15%. It is widening.
So the pattern is not “AI is coming for everyone.” It is much stranger and much more specific than that. Experienced people in exposed jobs are broadly fine. The squeeze is landing almost entirely on the people who have not started yet.
The Part Almost Nobody Quotes: The Researchers Disagree
Here is where it gets genuinely interesting, and where most coverage stops short.
The two Stanford teams do not agree on what is causing this. And both of them say, in writing, that they cannot prove it.
The problem is timing. The decline in young-worker hiring starts around 2022. But ChatGPT only launched in November 2022, and the tools were fairly limited then. The brief’s authors put it plainly: the timing of the impact is somewhat surprising, because AI capabilities were very limited in 2022.
Meanwhile something else happened first. The Federal Reserve began raising interest rates in March 2022 — months before the AI boom started. Higher borrowing costs make companies cautious, and cautious companies stop hiring juniors before they do anything else.
When additional controls for those factors were applied, the brief notes, employment declines among entry-level workers are not notable until 2024. Their honest summary: it seems plausible that factors other than AI are driving declines in hiring young workers around 2022.
The Digital Economy Lab answered that argument directly in February 2026. They found something unexpected — the jobs most exposed to AI are, on average, less sensitive to interest rates, not more. And the decline in AI-exposed occupations shows up in both the rate-sensitive half of the economy and the rate-insensitive half. Interest rates, they argue, do not explain why the drop is concentrated in AI-exposed entry-level roles specifically.
And still they add the caveat themselves: these are descriptive patterns, not causal estimates. They state outright that they do not view the work as definitive evidence of AI’s labour-market effects. They also note the gap shrinks once education is accounted for.
Two serious research teams, same university, same data era, opposite readings — and both of them refusing to overclaim. That is what a genuinely open question looks like. Anyone telling you with confidence that AI took your first job is telling you something the researchers themselves will not say.
The Detail That Changes How This Feels
Buried in the findings is one fact that ought to be the headline.
The adjustment is happening through hiring, not firing.
Nobody is being marched out of buildings. Layoffs are not driving this. Pay is not collapsing either — the changes show up in employment levels, not in what people are paid. What has happened is that a door stopped opening as often.
That is a completely different fact from the one the silence implies.
“You applied and were found wanting” and “the position was never opened” feel identical from the outside. They produce the same empty inbox. But they are not remotely the same event, and only one of them is about you.
The rung is not being sawn off. It is just not being built this year.
Where the Losses Are — and Where They Aren’t
This is the practical part, and it is the part worth acting on.
The declines are concentrated in occupations where AI automates the work — where a tool does the whole task instead of a person. In occupations where AI complements the work — where the tool makes a person faster at something a person still has to do — employment is flat or rising.
That distinction is doing enormous work, and it is almost never mentioned. Same technology. Opposite outcomes. The question is not “is my field exposed to AI” but “does AI replace this task or speed it up.”
It is also worth knowing that exposure lists are not forecasts. When a government classification put 206 jobs in the top AI exposure tier, the agency itself said it was not a prediction — and several of those jobs are projected to grow.
And the productivity revolution the whole panic rests on has not exactly arrived. When nearly 6,000 executives were surveyed about AI productivity gains, nine in ten reported none — and then predicted gains anyway.
Everyone is guessing. Including the people doing the hiring.
What a Stretch of Being Unneeded Actually Does
None of that changes the hardest part, which is not economic at all.
It is what an unwanted empty stretch does to how you see yourself. Work is how most people answer the question of whether they are useful. Take the work away and the question does not go away with it — it just sits there, unanswered, getting louder.
There is a very old idea, far older than any labour market, that runs directly against the panic in that question.
In the oldest stories people have kept and retold about being formed by God, almost nobody is formed while they are being used. They are formed in the gap. Tending animals in country nobody wanted. Sitting in a prison for something they did not do. Walking through wilderness for years with no title, no audience and no visible progress. The stretch that felt like being passed over turned out to be the stretch that made them into whoever they later became.
Not one of them enjoyed it. Not one of them could see it while it was happening. It is a strange pattern, and it shows up far too often to be a coincidence.
What You Can Actually Do This Week
None of this requires money, contacts, or a good week.
- Aim at complement, not automation. Look at the work you want and ask honestly: does a tool do this whole task now, or does a tool just make a person faster at it? Move toward the second. This is the single highest-value filter available right now.
- Make one small piece of real proof. Not a course. Not a certificate. One finished thing someone can look at — a fixed problem, a written analysis, a working sample. Entry-level hiring has become a proof problem, and proof is one of the few things you can build with no permission and no budget.
- Send fewer, better. Thirty applications into silence teaches you nothing and costs you a great deal. Five that a real person actually reads teaches you something either way.
- Change what the silence means. Write it down if you have to: the door did not open. That is not the same as being turned away. Say it as often as you need to, because the other version is far more expensive than it looks.
- Keep the shape of a week. Sleep, walk, eat, see one person. A stretch with no structure erodes people faster than the job hunt does.
If the harder question underneath all this is the one about being useful at all, our free What Is My Purpose? assessment is a decent place to think it through — it takes a few minutes and it does not ask for a CV.
The market will open again. It always has. Hiring is the first thing companies cut and among the first things they restore, and the people who kept building through the closed stretch are the ones standing closest to the door when it moves.
You are not being rejected thirty times. You are standing in front of a door that has not been opened yet — and that is a fact about the door.
Questions People Are Asking
Is AI causing unemployment among recent graduates?
Nobody has proven it. A July 2026 Stanford Institute for Economic Policy Research brief found that once additional controls were applied, employment declines among entry-level workers were not notable until 2024, and concluded it seems plausible that factors other than AI drove the decline in young-worker hiring around 2022 — pointing to interest rate rises that began in March 2022, months before ChatGPT launched. Stanford’s Digital Economy Lab disagrees, having found that AI-exposed occupations are on average less sensitive to interest rates, but describes its own findings as descriptive patterns rather than causal estimates. The question is genuinely unresolved.
What is the unemployment rate for recent graduates in 2026?
Unemployment among recent graduates in the United States reached 5.6% in early 2026, according to a Stanford Institute for Economic Policy Research policy brief published in July 2026. That figure was up 1.6 percentage points compared with three years earlier.
Are entry-level jobs actually disappearing?
Entry-level opportunities in AI-exposed occupations have contracted measurably. Stanford’s Digital Economy Lab reported in August 2026 that employment among workers aged 22 to 25 in highly AI-exposed occupations sits roughly 19% below where it would be had it kept pace with similarly aged workers in less-exposed occupations — a gap that widened from 15% in July 2025. Importantly, the adjustment is occurring through reduced hiring rather than through layoffs, and it is concentrated among early-career workers rather than experienced ones.
Which kinds of jobs are least affected by AI right now?
Employment declines are concentrated in occupations where AI automates a task outright. Occupations where AI complements human work — making a person faster at something a person still performs — show flat or rising employment. The useful test is not whether a field is described as AI-exposed, but whether the technology replaces the task or accelerates it.
Why is it so hard to get a first job with no experience?
Entry-level hiring is usually the first thing employers reduce when conditions tighten, because junior roles cost money to train before they return value. Current data shows the labour market adjusting through reduced hiring rather than layoffs, which means fewer openings rather than more dismissals. An empty inbox in that environment more often reflects a position that was never opened than a judgement made about the applicant.
What Do You Think?
If the entry-level rung really is being pulled up, whose job is it to rebuild it — employers, schools, governments, or the people trying to climb it? There is no obvious right answer, and reasonable people land in very different places. Tell us where you land in the comments.
Share This
Two Stanford teams looked at the same data on vanishing entry-level jobs and reached opposite conclusions about the cause. Both wrote down that they can’t prove it. Everyone quoting this with total confidence has read neither. https://bgodinspired.com/index.php/personal-growth-and-life-skills/why-entry-level-jobs-are-disappearing/
The most useful thing I’ve read about the job market this year: the change is happening through hiring, not firing. “You were rejected” and “the role was never opened” feel identical from the outside. They are not the same event, and only one of them is about you. https://bgodinspired.com/index.php/personal-growth-and-life-skills/why-entry-level-jobs-are-disappearing/
If you’re sending applications into silence right now, read this before you decide what it means about you. The numbers say something very different from what the empty inbox is telling you. https://bgodinspired.com/index.php/personal-growth-and-life-skills/why-entry-level-jobs-are-disappearing/
