AI is removing the first rung of the career ladder — and we have a responsibility to help fix that

After giving a keynote on AI and the future of work at the University of Greater Manchester, UK, I kept coming back to one fact about the institution itself. It traces its roots to a mechanics’ institute founded in Bolton in 1824.

That mattered because work was changing. New machinery was reshaping industry, established skills were losing value and people needed access to different knowledge. The response was not to pretend that technology could be held back. It was to invest in people so that more of them could participate in what came next.

Two hundred years later, we face a similar responsibility. This time, however, the machines do not only lift, cut or manufacture. They can write, analyze, code, summarize and make recommendations. They are reaching directly into the tasks that have traditionally formed the first years of a professional career.

That creates an uncomfortable question for CIOs. If AI removes the work through which inexperienced people became experienced, who develops the managers and leaders we will need next?

The first rung is disappearing

The pressure on young people is already significant. In the UK, 981,000 people aged 16 to 24 were not in education, employment or training between April and June 2026. That was a small quarterly improvement, but still 30,000 higher than a year earlier. Research cited by the Work Foundation suggests the number of starter jobs has fallen by 49% over the past decade.

The graduate market is intensely competitive too. Employers now receive around 140 applications for each graduate vacancy, compared with 38 two decades ago, while the Institute of Student Employers has forecast a further 7% fall in graduate vacancies in 2026.

AI is not the only reason. Economic growth is weak, employment costs have risen and many organizations are holding on to experienced people while delaying recruitment. Blaming every lost vacancy on generative AI would be lazy. Claiming it is having no effect would be equally difficult to defend.

Jobs are bundles of tasks. AI does not need to replace a whole job to change how many people are hired or what an entry-level employee is expected to contribute. First drafts, routine analysis, research summaries, document review, meeting notes, basic coding, testing and customer responses are all being automated or accelerated now.

Many of those tasks sit inside junior roles. They may be repetitive, but they have also been how people learned. A junior analyst cleans the data and notices that the numbers do not reconcile. A trainee reads the documents and begins to recognize risk. A developer fixes smaller defects and gradually understands why the architecture behaves as it does. Context and judgement grow through the work.

Updated research from Stanford’s Digital Economy Lab found that employment among US workers aged 22–25 in AI-exposed occupations stood 19% below where it would have been if it had kept pace with less-exposed occupations. The researchers are careful to describe this as an early, descriptive indicator rather than a causal estimate. They also found that the decline was concentrated in occupations where AI substituted for human tasks. Employment was flat or rising where AI complemented people.

That distinction should be receiving far more attention inside executive teams.

The short-term margin trap

I have previously argued that the CIO is becoming the most commercial role in the boardroom. Technology is now woven through almost every decision about customers, operations, people and growth. The best CIOs connect technology investment to revenue, margin, resilience and competitive advantage.

The same commercial discipline must apply to AI. Organizations are under genuine pressure to grow revenue, protect margin and control cost. If AI can reduce effort, improve service or increase capacity, leaders should use it. Preserving pointless work is not a talent strategy, and nobody needs to spend three years copying numbers between spreadsheets to build character.

But the short-term spreadsheet and the long-term capability of the organization are not always the same thing.

Removing junior roles may improve this year’s cost line. If those roles were also how people acquired the judgement needed to become senior analysts, engineers, lawyers or managers, the organization may simply have moved the cost and risk several years down the road. A business can protect today’s margin while weakening tomorrow’s resilience.

There is a better route. A large field study published by the US National Bureau of Economic Research found that generative AI increased productivity among customer support agents by 14% on average, with the greatest gains among less experienced workers. The technology helped newer people benefit from knowledge embedded in the organization and become effective more quickly.

That is a much more useful ambition than using AI simply to avoid hiring them. The question is not whether a task can be automated. It is whether automation removes waste, removes a learning opportunity or does both. Leaders need to know which before approving the business case.

CIOs need to rebuild the apprenticeship

As CIOs, we have spent years arguing that we are business leaders, not simply technology leaders. That comes with responsibility. We cannot encourage the organization to lean into AI, celebrate the immediate productivity and margin gains, then distance ourselves from what happens to the people and capabilities underneath them.

Our influence extends well beyond the technology function. We can model responsible adoption, help other executives understand where AI augments work and challenge business cases that depend entirely on removing headcount. We can ask what capability will be lost; how future experts will develop and whether the organization is retaining enough entry-level opportunity to sustain its leadership pipeline.

This does not mean protecting every existing graduate role. Many should change. Junior employees should use AI, and organizations should expect them to contribute more quickly because of it. But the role must still contain real problems, supervised responsibility, feedback and enough exposure to the underlying work to develop judgement.

The education sector has work to do as well. The 2026 HEPI Student Generative AI Survey found that 95% of UK undergraduates were using AI and 94% used it to support assessed work. Yet fewer than half believed teaching staff were helping them build the AI skills required for their careers. Student adoption has already happened. Structured capability has not caught up.

CIOs and employers can help close that gap. Universities need current examples of how tasks are changing, not a three-year-old view of a graduate job. Employers need assessments that reveal reasoning, verification and judgement rather than rewarding the most polished AI-assisted application. Both need to create opportunities for students to work with messy data, imperfect systems, competing priorities and actual consequences.

Within our own organizations, we should measure AI programmes against more than hours saved and roles removed. Are people becoming capable more quickly? Is knowledge spreading beyond a few experienced employees? Are junior colleagues taking on more valuable work? Are we building skills the organization will need in three, five and ten years?

This is a tough ask in the current global economy. Leaders face pressure to deliver growth now and protect margin now. Headcount reductions are immediate and easy to put into a business case. The value of a future manager who has not yet been hired is much harder to model.

That does not make the value less real.

Resilience is not confined to cyber controls, continuity plans and reliable platforms. It includes having people with the knowledge, confidence and experience to make good decisions when conditions change. An organization that automates its entry-level work without rebuilding the route to expertise is creating a dependency it may not recognize until the experienced people leave.

The mechanics’ institutes of the nineteenth century did not protect people from industrial technology. They created access to the knowledge needed to use it, improve it and build new opportunity around it.

AI gives us the chance to do the same. Used carefully, it can help graduates become useful sooner, widen access to expertise and allow people to take on more valuable work. But that outcome will not happen by accident.

If CIOs help remove the first rung of the career ladder, we also have a responsibility to help rebuild it.