The global youth career problem is bigger than AI

It is tempting to explain the difficulty young people have entering work as an AI story. It is a better story than the real one and it is mostly wrong.

The entry problem was already there. Automation of junior work is making it worse, which is a different claim and leads to different solutions.

What was already broken

Several things, and none of them are recent.

Employers ask for experience for entry-level roles, which is a loop nobody inside it can break.
Education is optimised for credentials rather than demonstrated capability, so graduates arrive with a certificate and no evidence.
Hiring filters on proxies, which favours candidates whose background matches the filter rather than those who can do the work.
Apprenticeship as a structure largely disappeared from knowledge work without anything replacing it.

The loop that predates automation: entry-level roles asking for experience, education producing credentials rather than evidence, and hiring filtered on proxies that favour whoever already had advantages. — None of this is recent, and none of it was caused by AI.

What AI adds on top

It removes a large share of the tasks that entry-level roles consisted of, which weakens the commercial case for the roles themselves.

It makes the credential problem worse, because a polished application now proves even less than it did, so employers fall back harder on proxies like university, network and prior title. Those proxies favour exactly the people who already had advantages.

And it removes the mechanism by which somebody without a network could prove themselves: do the unglamorous work well for a year and become undeniable. That route is narrower now.

Why framing this as an AI problem produces bad answers

If the diagnosis is AI, the response is AI training, and a lot of that is being delivered. It is not useless and it does not touch the actual problem, which is that young people cannot get into positions where capability can be built and seen.

Somebody who cannot get a first role does not need more tool training. They need a way to demonstrate capability that does not require having already held the job.

What actually helps

The interventions that work are unglamorous and all point at the same thing: making capability visible without requiring a prior job.

Assess capability directly in hiring rather than filtering on proxies. It is more work and it finds people the filters reject.
Create genuine entry structures where somebody produces real work under real correction, which is what apprenticeship was.
Value demonstrated work over credentials, and say so explicitly in job adverts so people know what to prepare.
For young people: build something real, publicly, however small. Evidence is the one thing that routes around a filter.

Making capability visible without requiring a prior job: real work done under real correction, and something built that somebody can examine directly. — Evidence is what routes around a filter built on prior titles.

Common questions

Is AI causing youth unemployment?

It is making an existing problem worse rather than creating it. Young people faced experience requirements for entry-level roles, credential-focused education and proxy-based hiring long before automation. AI removes many entry-level tasks and weakens the signals that let outsiders prove themselves, which compounds all three.

How can young people start a career when entry-level jobs need experience?

By producing evidence that does not depend on having held the job: something built and public, however small, a decision made and its outcome, a problem solved where the work can be examined. Evidence is what routes around a filter designed around prior titles.

What should employers do about entry-level hiring?

Assess capability directly rather than filtering on proxies, and build genuine entry structures where somebody does real work under real correction. Both cost more than reading CVs, and both reach people the filters currently reject.