As companies automate routine work, they’re increasingly choosing not to hire for the junior roles that once served as the starting point for a career. The consequences may be difficult to see in employment data, but could reshape the workforce for years to come. Nexpher Images via Getty Images

On Oct. 22, Barclays announced plans to cut around 5,000 roles, equivalent to roughly five percent of its workforce and concentrated largely in operational and functional jobs, as the British bank scales A.I. to boost efficiencies and slash costs. Chief executive C.S. Venkatakrishnan described A.I.’s role as “the icing on the cake, not the cake.”

The phrasing is revealing. A.I. may be accelerating a cost-cutting decision that would have been difficult to justify on its own. But the icing on the cake is this: historically, reducing headcount meant reducing a company’s capacity to deliver. A.I. is changing that equation. As it takes on more of the coordination and process work once handled by large teams, businesses can reduce staff without necessarily reducing output. The remaining employees take on the work that still requires human judgment while A.I. absorbs more of the rest. 

My money would be on this: they saw that A.I. was capable of handling much of the orchestration work that was usually covered by many people. They tested it, and realized they could remove roles, absorb the savings on the wage bill and not lose any capacity in delivery or output. They’re a bank. They probably did very clever math. If automated systems can handle a significant share of the work previously performed by a team, how many people does the business still need? That question is already shaping hiring decisions, particularly at the junior level. 

Conversations with business leaders often begin with a version of the same dilemma. A company has automated a portion of its work, so what happens to the junior employees who once performed it? Some employers are investing in training those employees to take on mid-level responsibilities, a process that requires time and money. Others are letting junior staff go. And some are making the conscious decision to stop hiring entry-level employees altogether, even as existing junior staff help build the automations that make those roles less necessary. In effect, the people at the bottom of the career ladder are helping remove the rung behind them. 

Some roles are being closed, some aren’t being backfilled and others are being moved internally to mid-level. Which, given A.I.’s capacity to perform knowledge work, means those mid-level workers will be just as exposed in a few years as they were as junior employees.

A professional services firm I advise recently illustrated the tension. After deploying a substantial number of A.I. workflows, its global leadership directed the U.K. business to “cut the dead weight.” The—quite callous—instruction reflected a new reality for the firm: it no longer needed as many junior employees to complete routine work. But a reduced need for certain tasks does not make the people who performed them expendable. 

The firm’s U.K. managing director agreed that affected employees should be supported and reskilled where possible, and that any redundancies should be handled with care. The business is developing a transition plan, but it does not expect to continue hiring juniors. As employees leave, existing mid-level staff will absorb more of the work. That may solve an immediate staffing problem. It also raises the question of where the next generation of experienced employees will come from if businesses stop giving people the chance to gain experience in the first place. 

Some of the current pressure on employment is cyclical: boom and bust, boom and bust. Internationally, businesses are still carrying the weight of post-pandemic over-hiring, while higher costs and economic uncertainty are making employers more cautious. Yale’s Budget Lab has found little evidence so far of A.I. disrupting U.S. employment overall. Its executive director has also suggested that some CEOs may be attributing job cuts to A.I. when other pressures, including tariffs and interest rates, are at play. 

A.I. is arriving at a moment when the economics of hiring are already shifting. Many employers are paying more to have staff in situ than they ever used to. In the U.K., employer National Insurance, the tax businesses pay on top of employees’ salaries, increased from 13.8 percent to 15 percent in April 2025, while the salary threshold at which it applies fell from £9,100 to £5,000. For employers weighing the costs of adding an entry-level employee, automation offers another way to contain expenses. If a business can replace a portion of junior-level work with an A.I. subscription costing a few hundred dollars a month, the financial case can be difficult to ignore. 

The risk is that these decisions become permanent. When an economy stabilizes and returns to growth, businesses are unlikely to restore roles that automation has made unnecessary. A job that disappears because demand has fallen might return when demand picks up. A job that disappears because its core tasks have been automated will not. 

The consequences extend past the employees who lose their first jobs. If you’ve never been a junior something, you can never be a senior anything. Entry-level work is often where people learn how an organization operates, from how to research a problem, build an argument, check an output, manage a client relationship and understand the decisions behind a finished piece of work. These tasks can be repetitive, but they are also how professional judgment develops. A graduate who never does the grunt work may struggle to understand, evaluate or improve the work that senior employees deliver. 

That familiar idea that no one becomes a senior professional without first being a junior one is more than a matter of career progression. It describes how expertise is built. If businesses remove the early-career work that once served as training, they risk weakening the pipeline of people capable of taking on more complex responsibilities. The short-term result may be higher margins and lower headcount. The longer-term result could be a workforce with fewer opportunities to develop the experience its employers will eventually need, resulting in workers who end up in a state of professional atrophy. 

Middle management faces a related pressure. Much of the role involves coordinating work such as scheduling meetings, managing correspondence, preparing presentations and building or analyzing spreadsheets. A.I. tools are often capable of handling these functions. As they take over routine coordination, teams can become more productive with fewer people. But productivity gains do not automatically create more work. If a team of 10 can now produce what previously required 20, a business may have little reason to maintain the larger team. Over time, the middle rung of the organizational ladder can thin out even as the entry-level pipeline narrows. 

Decisions typically move faster than the employment data shows. Data can trail decisions by years. Hiring decisions often change before the consequences show up in workforce statistics, and official figures capture jobs that exist, not roles that employers have internally decided never to create. Quarterly employment data is a snapshot of past decisions, not necessarily a guide to what companies are planning next. 

There is evidence that younger workers in exposed occupations are already feeling the pressure. Stanford’s Digital Economy Lab found in August that employment among 22- to 25-year-olds in the jobs most exposed to A.I. was 19 percent below that of their peers, up from 13 percent a year earlier. The gap was driven largely by fewer new hires. That finding points to a shift in the entry-level labor market, even as broader employment data has yet to show widespread disruption. 

But the more difficult trend to measure is the one that never appears in a layoff announcement. The graduate role is not advertised, the junior analyst who is never hired, the assistant position quietly removed from a budget. Those decisions may not register as job losses, but collectively they could reshape the labor market just as profoundly. A.I. does not have to eliminate entire professions to alter the structure of work. It only has to make businesses less likely to hire people at the bottom of the ladder, and, eventually, less reliant on the layers above them. The jobs A.I. is taking may be the ones that never get posted. 

Chris Simmance is the founder of Orca and the author of The Next Rung, published in paperback by Practical Inspiration Publishing on Jan. 19, 2027.