The tension: “entry-level” no longer means entry

Look at any job board and you’ll still find the listings: Marketing Assistant, Junior Analyst, Clinic Receptionist, Graduate Scheme. The titles survive. What hasn’t is the part where a beginner could actually get one.

In the UK, the squeeze is particularly visible at the macro level. Almost one million young people are classed as “NEET”, meaning not in education, employment, or training, and the number actively seeking work spiked 12% in a single quarter. Rising employment costs and AI eroding entry-level roles have pushed youth unemployment to an eleven-year high. The government’s response was to issue a “youth guarantee” of six months’ paid work for long-term unemployed 18-to-21-year-olds. But this is a floor under the problem, not a fix for it.

The mechanics underneath are simple and brutal. AI tools let candidates apply in seconds, so postings drown in volume. Employers cope by demanding more: extra credentials, prior experience, professional polish straight out of university. The “entry-level” role quietly acquires a two-to-three-year experience requirement, and the beginner it was named for can’t get through the door. They need a job to get experience and experience to get a job.

And that’s just the surface. The more interesting question is why the rung is being removed in the first place, because the answer isn’t a hiring fad but a structural one.

Why it’s happening: a barbell and a broken signal

Two forces are hitting the bottom of the labor market at once. The first is moving the jobs themselves, automating the work that beginners used to cut their teeth on. The second is wrecking the signals employers relied on to spot a promising beginner in the first place. One removes the rung; the other removes your ability to tell who could have climbed it. That combination is why the first rung breaks before any other.

Force one: the workforce barbell

For more than a decade, economists have tracked a hollowing-out of the middle of the labor market. The work of David Autor and others documented a U-shaped pattern: demand grows at the high-skill, high-wage end and at the low-wage service end, while routine middle-skill roles shrink. We call it “the barbell”, a weight stacked at both ends, the bar in the middle running thin.

What’s different now is what AI can reach. Earlier automation took routine manual work. AI takes routine cognitive work too: admin, coordination, basic analysis, first-draft reporting, the entry-level tasks that taught beginners the ropes while they earned. The rung isn’t disappearing by accident. It’s being automated.

The data points in the same direction. The World Economic Forum’s Future of Jobs Report 2025 finds employers expect to cut roles where skills have become obsolete and hire where new ones are needed, with 40% planning to reduce headcount as AI automates tasks and 70% planning to hire for new, more advanced skill sets. Companies are over-hiring at the top for people who can design systems and handle ambiguity, and leaning on automation for what used to be mid- to low-level knowledge work.

That guts the traditional progression path. The junior-to-mid-to-senior ladder assumes a populated middle to climb into. Automate the middle, and the bottom rung has nowhere to lead.

Force two: the old system has collapsed

The second force is quieter but just as decisive. The signals employers used to lean on to determine promise in a beginner have stopped working.

In a pre-ChatGPT world, a polished resume, a tailored cover letter, and a well-written application email genuinely signalled something: effort, motivation, a baseline of competence. Generative AI made all of that cheap and infinite. Candidates use AI to produce flawless prose. Employers use AI to scan it. Everyone optimizes for keywords. Very little of what’s exchanged reflects what a person can actually do.

The proxies are eroding across the board. Written materials are no longer trustworthy as evidence of effort or skill. Keyword-based screening is trivially gamed by anyone with a chatbot. And degree requirements are being quietly downgraded: PwC's 2025 AI Jobs Barometer found degree requirements fell by 7 to 9 percentage points between 2019 and 2024 for AI-exposed jobs as employers shifted toward skills. The shift runs deep enough that, in our recent State of Hiring for AI Fluency report, 53% of hiring managers said they now prefer a candidate with strong AI fluency over one with deep domain expertise. 

This is why “more experience” became the default filter. When you can’t trust the signal, you reach for the bluntest available proxy, years served. That proxy is exactly the one a beginner can’t produce.

It’s not just the resume that’s broken; the entire signalling system of hiring has collapsed. Companies are responding by hiring fewer people and putting every role under more pressure to prove its value. An analysis by PwC of US data shows AI-exposed entry-level roles are seven times more likely to require traditionally senior-level skills such as judgement and leadership.

The UK as the live case

Nowhere does the barbell show more clearly than in UK immigration policy. The decision to raise the Skilled Worker visa salary threshold effectively priced out mid-skill sponsorship. Companies can now realistically sponsor only senior or shortage-skill roles. The mid-tier sponsored hire, the classic rung-two job, has been legislated out of reach. 

Layer that onto a near-million NEET population and an eleven-year high in youth unemployment, and the UK becomes the clearest case study of what happens when the middle hollows and the signals fail at the same time. The jobs concentrate at the top. The beginners pile up at the bottom. And nothing connects the two.

Why employers should care: this is your pipeline

It’s tempting to read all of this as a young-person problem. It isn’t. An employer who stops hiring beginners is quietly dismantling their own future workforce, and several costs land sooner than expected.

There’s a deeper point here. The barbell tells you where the work is going. It doesn’t tell you the talent has vanished. The people who can learn fast, adapt to new tools, and grow into the high-autonomy roles that are multiplying at the top of the barbell are out there right now, sitting in that NEET statistic, sitting in the application pile you’re filtering by years of experience. The pipeline problem and the hiring problem are the same. You can’t build the top of the barbell if you’ve stopped hiring anyone capable of climbing toward it.

Which raises the real question. If experience and credentials no longer signal capability, and the routine tasks that used to ease beginners in are being automated, what should you be hiring for?

The fix: forget confidence, assess for capability

Here’s the response most employers have: when the old signals stop working, they don’t replace them, they just demand more of them. More experience, more credentials, more polish. They tighten the filter on exactly the proxies that broke.

The way out isn’t a better proxy but a different kind of evidence.

Learn from the AI hiring mess

You don't have to theorize about what happens when employers hire for the wrong signal, because it's already played out in AI hiring. Employers know they want AI fluency, yet 59% report making a bad AI hire, i.e., someone who sounded capable in the interview but couldn't apply the skill on the job. 

The reason is almost always the same. They set the bar at what a candidate can describe, not what they can do, and an interview rewards the person who talks about the work most confidently, not the one most able to perform it. We’ve labeled this the “confidence vs competence” problem, and we've unpacked it in full, along with the framework that fixes it.

It's the exact mistake employers are about to repeat with entry-level hiring. When you hire for confidence, you get storytellers. When you assess for competence, you get the hire.

The same principle rebuilds the first rung

Map that back onto entry-level. The barbell removed the routine tasks that let you watch a beginner learn on the job. The signalling collapse means resumes and confident interviews tell you nothing reliable. So the answer is the same as it is for AI hiring: stop reading proxies, start generating evidence.

It also means rethinking what counts as a qualification in the first place. Speaking at the 2026 GenAI Summit, TestGorilla VP of Engineering Chris Newton argued that long tenure in a single technology, the classic senior-level proxy, is losing its meaning: "this person's been doing Java for 20 years... that for me becomes less important if they've got the core principles in play." 

He reasons that AI has "lowered the barrier" to working in adjacent skills, so the candidate who can learn and move across domains now outvalues the one with the longest track record in just one. For entry-level hiring, the beginner you're filtering out for lack of experience may be exactly the fast-learning generalist the top of the barbell is crying out for.

For a beginner, that means assessing the things that actually predict whether they'll grow into the role.

This is what skills-based hiring actually delivers, and it’s worth being precise about why it solves this problem specifically. It widens the pool past credentials to career switchers and people who built skills through alternative routes. It surfaces potential the resume hides. And because you’ve seen what a person can do before you hire them, it gives you the confidence to invest in paid apprenticeships and internships, providing experience rather than demanding it.

Where policy is already pointing

The UK’s own policy direction backs this up. The new Level 4 AI and Automation Practitioner apprenticeship launched by Skills England is, in effect, a state-built first rung for the AI era, a structured on-ramp that trains beginners for high-autonomy work rather than demanding they already have it. We believe it’s the right move, with one condition: it has to train and assess for behavioral fluency and judgment, not just tool familiarity. The same warning applies to every employer rebuilding their own entry path. An apprenticeship or a graduate scheme is only as good as what it assesses for. Build it around demonstrated capability and it produces the talent you can’t currently hire. Build it around the old proxies and you’ve recreated the problem with a new label.

How TestGorilla rebuilds the step on the ladder

This is the gap TestGorilla was built to close. Our entire platform runs on one principle: hire for what people can do, not what they claim.

Our library of 350+ science-backed assessments lets you test the things that actually predict entry-level success, cognitive ability, problem-solving, role-relevant skills, and the learning agility that determines who grows. You can pair skills tests with AI-driven interviews and structured scoring so that every candidate is evaluated on the same evidence, not on who interviews most confidently. No vibe checks. No storytellers. Just a clear, auditable signal of capability you can act on.

That’s how you rebuild the first rung in a labor market that’s trying to remove it. The barbell isn’t going away, and the old signals aren’t coming back. But the talent is still there. The only question is whether your hiring process is built to find it.

Stop hiring for the polish. Start assessing for the potential. Create your free TestGorilla account and build a ladder people can climb.