The Missing Rung on the Career Ladder
On Sunday, the UK woke to a number that reads less like a statistic and more like a weather report: a 27% chance of losing your job to AI within five years. That figure—lifted from Randstad’s 2026 Workmonitor and carried by the Guardian—doesn’t forecast a single storm. It sketches a pressure system building across offices and high streets, one where confidence, investment, and anxiety collide.
The tension is not about whether AI works. In the UK, two-thirds of employers say they poured money into AI over the last year, and over half of workers report their companies are nudging them to use the tools. Many already do: a majority say AI has made them more productive. Yet nearly half believe the gains will flow upstairs, not across the payroll. That belief, however accurate in any one firm, shapes behavior. When workers expect a one-way transfer of value, productivity becomes a story about extraction rather than empowerment.
Look closer and the anxiety concentrates where it hurts the engine of social mobility: at the first rung. Gen Z and early-career workers see managers swapping “learn by doing” for “automate the doing.” The entry-level spreadsheet, the first customer ticket, the preliminary draft—tasks that once taught judgment in small doses—are being routed to systems marketed as tireless and impartial. Baby boomers, by contrast, report more confidence, perhaps because they are the ones deciding which tasks become prompts and which remain apprenticeships.
This is the AI reality gap. Employers are accelerating adoption and relabeling work at scale; employees experience both convenience and precarity. The productivity story is not fiction. At the task level, AI is shaving minutes and improving outputs. But careers are not a sum of isolated tasks. They are sequences—time spent acquiring tacit knowledge, building context, and earning responsibility. Remove low-complexity work wholesale and you remove practice reps. The ladder doesn’t vanish; it becomes a shelf out of reach.
From operators to orchestrators
Randstad’s dataset extends far beyond the UK: 27,000 workers, 1,225 employers, 35 markets, and an analysis of three million job postings. Four in five workers expect AI to change their daily tasks. The most striking market signal, though, is the rise of “AI agent” roles—skills linked to building, overseeing, and coordinating software agents jumped 1,587% through 2025. That growth is not a side note; it’s a reorganization of the firm.
Companies are quietly adding a new layer of work: orchestration. Instead of doing the task, you specify it, monitor the agent, integrate its output with other systems, and troubleshoot edge cases. In theory, this creates new jobs. In practice, it creates a pipeline problem. Orchestrators need judgment formed by hands-on exposure. If automation absorbs the tedious but formative tasks, where do future orchestrators come from? Without deliberate redesign, the labor market will try to mint pilots who never learned to taxi.
Why the UK feels like a flashpoint
The UK’s service-heavy economy is dense with the transactional work AI now targets: customer support, claims processing, marketing ops, back-office finance, policy drafting. It’s also a context where productivity growth has been a political obsession and margin pressure is acute. That combination invites rapid deployment. If you believe the numbers, British employers are indeed moving faster than their workers’ trust can keep pace.
The result is a subtle shift in social contracts. For decades, the implicit bargain was that entry-level toil would be exchanged for training and trajectories. The new bargain is unsettled. Workers see AI accelerating, but the pathways to absorb its benefits—reskilling, new internal ladders, shared productivity dividends—are foggy. As one executive voice put it, “AI is not a rival to labour; it should be seen as key to augmenting tasks.” That is less a reassurance than an instruction manual. Augmentation only feels like augmentation if someone invests in the human half of the loop.
Designing fairness into the deployment curve
This moment doesn’t require sentimentality about pre-AI workflows. It requires design. If firms are serious about orchestration at scale, they need to manufacture the missing rungs: simulated environments where juniors rehearse complex judgment with agents; rotations that pair human operators and agent supervisors; hiring metrics that reward teams for progressing early-career staff into orchestration roles. If productivity jumps, codify how the gains flow—into wages, into training budgets pegged to automation savings, into shared time dividends that let teams learn new tools without working two jobs.
Unions and policymakers have their own levers: modernizing apprenticeship frameworks around AI-first work, setting transparency norms for automation impact assessments, and making training a precondition—not an afterthought—of large-scale deployments. Inside firms, governance can get specific: define which tasks are routed to agents and why; publish promotion criteria that value agent management as a teachable craft, not an innate talent discovered late in a career.
Stripped to its core, yesterday’s headline is not about which jobs disappear. It’s about whether the next generation gets a fair shot at building mastery in a workplace that increasingly runs on prompts, policies, and pipelines. The anxiety is rational. So is the opportunity. If employers want trust, they will have to build it like any other capability: with budgets, with measurements, and with the humility to keep a rung within reach for those expected to climb.
