AI Erases the On-Ramp: Entry-Level Jobs Now Require Senior Skills

Here is what PwC's global workforce leader said about what AI is doing to entry-level jobs: "AI is removing some of the routine work that once acted as an apprenticeship."

That sentence deserves to be read slowly. The routine work was not just filler. It was the training ground. It was how people learned to exercise judgment before they were trusted with it. Remove the routine work, and you remove the process by which junior employees became senior ones. What you are left with is a job market that wants the output of experience without providing the conditions to develop it.

Key Points

• The skills gap is quantified: Entry-level roles most exposed to AI are now seven times more likely to require senior-level skills — judgment, leadership, empathy — than the least AI-exposed entry-level roles, according to a PwC analysis of over one billion job ads across 27 countries.
• The growth gap is widening: AI-exposed entry-level roles that added higher-level requirements grew 35% from 2019 to 2025. Comparable roles that did not add those requirements shrank 10%.
• Early-career hiring is contracting: PwC itself plans to cut U.S. entry-level hiring by approximately a third over three years, reducing the offices where new consultants can work from 72 to 13.
• Recent graduates are worse off than the average worker: The Federal Reserve Bank of New York reports that recent U.S. college graduates are now more likely to be unemployed than the general workforce, reversing a historical pattern.
• Companies using AI are adding workers overall — just not entry-level ones: Jobs at the most AI-exposed firms grew 52% since 2018 versus 36% at the least-exposed, but the growth is concentrated in roles that already require advanced skills.

Here is what PwC's global workforce leader said about what AI is doing to entry-level jobs: "AI is removing some of the routine work that once acted as an apprenticeship."

That sentence deserves to be read slowly. The routine work was not just filler. It was the training ground. It was how people learned to exercise judgment before they were trusted with it. Remove the routine work, and you remove the process by which junior employees became senior ones. What you are left with is a job market that wants the output of experience without providing the conditions to develop it.

What the PwC Data Shows

PwC analyzed more than a billion job ads in 27 countries, including 2.4 million entry-level roles in the United States. The findings describe a workforce splitting into two tracks with very different trajectories.

The first track is what PwC calls "professionalized" roles — radiologists, recruiters, analysts — where AI handles the routine processing and humans are retained for judgment. These roles account for roughly 22% of advertised jobs. They are growing twice as fast as the second track and have posted 42% faster wage growth since 2021.

The second track is "democratized" roles — IT service management and similar functions — where AI makes the job accessible to non-experts. These account for roughly 52% of advertised jobs. They are growing more slowly, paying less, and represent the category most at risk of continued compression.

The most AI-exposed roles across both tracks are adding human-intensive requirements — empathy, creativity, judgment — at 2.5 times the rate of the least AI-exposed roles. The message to job seekers is becoming impossible to misread: if your role touches AI, you are expected to bring the things AI cannot replicate, and you are expected to bring them from day one.

The Apprenticeship Model Is the Casualty

The structural problem here is not automation in the sense that was originally feared. Widespread job elimination is not what the data shows. What it shows is subtler and in some ways more damaging to the pipeline of human expertise: the elimination of the learning curve.

Every senior professional with good judgment developed that judgment somewhere. Usually it developed through years of handling routine tasks, making low-stakes mistakes, watching how more experienced colleagues approached problems, and building pattern recognition through repetition. That process had a name. It was called an entry-level job.

When AI absorbs the routine work, it also absorbs the training mechanism embedded in that work. The companies benefiting most from AI are the ones that can immediately deploy the judgment and creativity AI cannot replicate. They are hiring people who already have it. They are not investing in the infrastructure to develop it in people who do not.

PwC's own decision to cut entry-level hiring by a third and reduce its new-hire office footprint from 72 locations to 13 is not an outlier. It is the leading indicator.

Who Bears the Cost

The Federal Reserve Bank of New York's finding that recent college graduates are now more likely to be unemployed than the average worker is the human consequence of this structural shift, and it should not be absorbed as a data point without registering what it represents.

An entire generation entered a credential system that promised access to professional opportunity and is finding the entry points have narrowed significantly, precisely as that credential became more expensive and more years of their life were spent acquiring it. The jobs that were supposed to be the first step are now asking for skills that typically take years of on-the-job development to build.

The companies best positioned in an AI economy are capturing real gains. Jobs at the most AI-exposed firms grew 52% since 2018. Wages at those firms rose 24% versus 17% at the least-exposed. The productivity gains are real. The distribution of those gains is not reaching the people entering the workforce right now.

What This Requires of Every Organization

For business leaders and growth teams, the PwC findings carry operational implications that go beyond workforce sympathy. The companies winning in an AI-augmented market are the ones that invested in human expertise before AI arrived, giving them a talent base capable of handling the judgment layer AI left behind. That advantage does not reproduce automatically. It requires deliberate investment in how people develop skills, and it requires organizations to think seriously about whether they are contributing to or drawing down the pipeline of human expertise in their industries.

Marketing and growth functions are not exempt from this dynamic. The most AI-exposed marketing roles — content strategy, campaign analysis, performance marketing — are exactly the ones PwC's framework would classify as "professionalized." AI handles the production. Humans are retained for strategic judgment. But that judgment has to come from somewhere, and if the entry-level roles that developed it are being cut, the organizations making those cuts are borrowing against their own future capability.

Understanding how AI changes the human investment required to build a capable team is increasingly part of how growth strategy gets built. The productivity gains from AI are real. So is the cost of dismantling the systems that produce the human expertise those gains depend on.

If your organization is considering how to build AI-assisted marketing capabilities without hollowing out the team underneath, our growth experts can help you determine what that actually requires.