A new survey of 750 HR leaders across the United States, the United Kingdom, and India, run by Cognizant and Pearson, puts a number on something many managers already sense: AI is quietly taking over the jobs new hires used to do first.
On average, a third of entry-level tasks are now completed by AI. In the UK, that figure is close to 39 percent. These were often the boring, repetitive jobs nobody loved, but they were also how junior employees learned the business and earned trust before moving up.
That matters more than it sounds. Almost every HR leader in the survey, 96 percent, expects entry-level jobs to turn into AI supervision roles within five years. But less than half of companies are proactively training people for that shift, according to the same survey.
The bigger gap sits one level up. Ninety-five percent of HR leaders say middle managers are the most important factor in getting AI used correctly day to day. Yet training for those managers is largely missing too. Companies are handing managers a new job, overseeing people and AI systems together, without teaching them how to do it.
This is not an isolated finding. A Stanford study published last year tracked workers between the ages of 22 and 25 and found a real drop in employment specifically in the jobs most exposed to AI. Separate hiring data from the venture firm SignalFire shows new graduate hiring at major tech companies has fallen sharply since 2019.
Not everyone agrees AI is the main driver. A 2026 research paper from economists at Warwick, the London School of Economics, and Oxford tracked hundreds of millions of job postings and hires, and argues that remote work, not AI, deserves more of the blame. Their reasoning: supervising and training a junior employee is harder over video calls, so many companies simply hired more experienced people instead once offices went remote and stayed that way. Either explanation points to the same outcome, fewer junior hires learning the trade the old way.
The part business leaders should actually worry about is not the entry-level squeeze itself. It is what happens five or ten years from now when there are far fewer people who came up through the old system of making mistakes and getting corrected. Companies that skip building a replacement path for that experience will run out of qualified internal candidates for leadership roles.
The survey also found a shift in what companies now look for in future leaders. Sixty-seven percent say they value liberal arts backgrounds, meaning broad thinking and communication skills, more than they used to. And 64 percent now rate the ability to spot new problems above the ability to solve familiar ones.
None of this requires a dramatic response. It requires a plan. Companies that build a deliberate system for teaching judgment, giving junior employees real decisions to make under supervision rather than just tasks to complete, will have a deeper bench of capable managers in a few years. Companies that just let AI absorb the entry-level workload without replacing what it taught will eventually feel the gap at the top.