Two new surveys landed within days of each other this month, and read together they describe employers pulling up the ladder on entry-level talent while telling themselves the ladder was never that important. That is a mistake HR will be paying for within three years, and the data these same employers collected is already showing the bill.
The Open University’s Business Barometer 2026, a survey of 1,500 UK business leaders and 1,000 young people classified as NEET (not in education, employment, or training), found that 51% of employers say AI is changing how they hire. Nineteen percent have reduced entry-level recruitment outright, and 42% of that group point to AI adoption as the reason. At the same time, a separate survey of 500 mid-market decision-makers by the AI consultancy Klarus found 45% saying AI helps junior staff work better or faster, and 24% saying AI is creating new roles and opportunities rather than closing them off.
The counter-argument, stated plainly
That second finding is the honest case against what I am about to argue, and it deserves to be stated rather than waved away. If AI genuinely makes junior employees more capable faster, and generates new categories of work rather than only eliminating old ones, then cutting entry-level headcount is not employers pulling up a ladder. It is employers right-sizing a function technology has made more efficient, the same way spreadsheets thinned out armies of manual bookkeepers without permanently damaging accounting. Under that reading, fewer junior hires plus AI-assisted juniors doing more per head is normal productivity growth, and HR leaders who resist it are protecting headcount for its own sake.
Why I think that reading is wrong
The problem is that the Open University’s own data undercuts the efficiency story it is supposedly telling. Seventy-six percent of the employers surveyed said economic uncertainty, not AI capability, had made recruitment or training harder over the past year, and 43% said they had hired fewer staff overall in that period. That is a company pulling back broadly under financial pressure, then pointing to AI as the explanation for one specific cut. More damning: 57% of those same employers report an active skills shortage right now, and more than half say that shortage is already hurting organizational performance, with 54% citing increased workloads on remaining staff, 44% citing lower morale, and 35% citing declining wellbeing. An employer cannot simultaneously claim AI has made entry-level talent less necessary and report that it cannot find enough skilled people to do the work. Those two claims describe two different companies, and only one of them is telling the truth about what is actually happening on the floor.
What is actually happening, I think, is simpler and less flattering than an efficiency story. Junior roles have always been where a workforce trains its own replacement pipeline: the associate who learns judgment by doing the grunt work under a senior person’s correction, the analyst who builds pattern recognition by running the numbers wrong before getting them right. Klarus’s own research shows why that pipeline is still needed. Alper Gunaydin, chief technology officer at Klarus, put it bluntly when describing why so many mid-market AI pilots stall before reaching production: “Mid-market companies have a real advantage because they can often move fast, particularly when it comes to technology transformation… However, our research shows that too many pilots stall because companies lack AI expertise, quality data and effective governance.” That expertise has to come from somewhere. If it is not grown internally through entry-level roles, on the job, it has to be bought externally at a premium, later, from a shrinking pool of people who never got the entry-level rung to stand on.
What it means for the HR leader
Cutting entry-level recruitment because AI can absorb some junior tasks trades a visible cost today, in headcount, for an invisible cost later, in the internal supply of judgment the organization will need once the easy AI gains are used up. The Open University’s own respondents are already reporting that invisible cost arriving early, as skills shortages hurting performance now, not in some hypothetical future. Forty-two percent creating new roles is real, but it does not cancel out 19% reducing entry-level recruitment when the same employers cannot staff the skills they already need.
The honest move for HR leaders is not to freeze entry-level hiring and blame AI, nor to ignore AI’s genuine effect on junior workloads. It is to treat entry-level roles as the training infrastructure they have always been, and redesign what a junior employee spends their first eighteen months doing so judgment, not task volume, is what gets built. Employers who use AI as cover to skip that investment are not becoming leaner. They are quietly deferring a skills bill to a version of the labor market that will have far fewer discounted candidates left to pay it with.
The same screening tools reshaping who gets seen in the first place make this harder to fix after the fact. HRTech Edition has reported that AI recruitment tools are already screening out the skills HR says it wants, and separate reporting on the trust gap in AI hiring found candidates and employers reading the same automated decisions very differently. An entry-level pipeline that is both smaller and being filtered by tools nobody fully trusts is not a pipeline HR can afford to stop investing in.
