by Claude Opus 5.5

How might AI change the value of experience and the structure of career ladders (especially the availability of entry-level “learning tasks”)?

AI lowers the value of experience that consists mainly of having done routine tasks many times, and raises the value of judgement, context and accountability. The bigger change is structural. The junior tasks AI does best, such as first drafts, reconciliations, document review and basic code, were also how people built that judgement. Career ladders are therefore at risk of losing their bottom rungs just as the middle rungs become more demanding. UK data in 2026 show early signs in falling entry-level hiring in exposed occupations, and US data show “seniorised” junior roles. Cost pressures explain a good part of the decline.

Two kinds of experience, two different fates

Experience has always bundled two things. One is a pattern library, the accumulated familiarity that comes from handling hundreds of similar cases. The other is judgement and trust: knowing what matters, spotting what is wrong, managing a client and being the person whose name goes on the work.

AI is good at the first and poor at the second. In Brynjolfsson, Li and Raymond’s study of 5,179 customer-support agents, an AI assistant raised productivity by 14% on average and by 34% for novices, with minimal effect on the most experienced. The tool effectively handed newcomers the best performers’ accumulated patterns. On many measures, agents with two months’ tenure and AI performed as well as or better than agents with more than six months’ tenure without it. In Dell’Acqua and colleagues’ consulting experiment, below-average performers gained 43% and above-average ones 17%.

So AI flattens the experience curve for routine work. But the same study found that consultants who leaned on AI for a task beyond its competence were 19 percentage points less likely to get it right. Knowing where the tool fails is a form of experience that becomes more valuable.

Autor and Thompson’s research on expertise adds a further twist. When automation strips out the less expert tasks in a job, what remains is more specialised, so wages rise and employment falls. That describes what is happening to many professional ladders: the junior work is automated, and the remaining roles demand more.

The apprenticeship problem

Professional careers in Britain were built on a bargain. Juniors do high-volume, lower-risk work that is useful to the firm and trains them at the same time. Seniors supply judgement and client relationships. The trainee solicitor reviewing documents, the audit associate testing samples and the graduate developer fixing small bugs were all being paid to learn.

AI attacks that bargain from both ends. It makes the junior work cheaper to do without juniors, and it lets seniors do more of it themselves. The Governor of the Bank of England put the concern directly in December 2025: “what is it doing to the pipeline of people?” The risk is not only fewer graduate jobs now. It is fewer competent mid-career professionals in five to eight years.

What 2026 UK data show

The evidence is consistent with the bottom rungs thinning, especially in information-processing work:

  • Entry-level hiring by occupation. The DSIT and LinkedIn snapshot (June 2026) found 30 of 38 tracked entry-level occupations shrinking in the year to April 2026. Falls were steepest for accountants (−29%), graphic designers (−28%) and software engineers (−27%), while sales and customer-facing roles grew. It reports employers “favouring experienced candidates who already have the operational skills most entry-level candidates lack”.

  • Seniorised junior roles. PwC’s analysis of 2.4 million US entry-level roles finds the most AI-exposed are now seven times more likely to require traditionally senior skills such as leadership, and that these roles grew 35% since 2019. That is US evidence, not a UK finding, but it suggests the entry rung is being raised, not just narrowed.

  • The profession pipelines. The Big Four’s combined UK graduate and apprentice intake fell from 6,500 in 2023 to 5,400 in 2025, while Grant Thornton UK raised its intake 30% to 340 (as reported). Legal Cheek’s September 2026 count of training contracts across more than 100 law firms showed a 1.7% fall, with Linklaters reportedly going from 100 to 60. Legal Cheek attributes the fall mainly to solicitor apprenticeships, with AI only a “potential” factor.

  • The second rung. Morgan Stanley’s survey found UK AI-related cuts concentrated in roles needing two to five years’ experience (as reported). That suggests the squeeze reaches past the first job to the stage where people were becoming useful.

  • Readiness of new hires. In ISE’s 2026 development survey, 29% of employers report rising performance issues among new hires, up from 12% in 2022. 54% worry about AI’s effect on new hires, and typical development budgets are down 10%.

  • Employer attributions. In a Work Foundation survey reported in August 2026, 36% of businesses had cut entry-level vacancies in the past year. Among large firms that cut, 60% blamed AI or automation, against 25% of small firms (as reported).

The cost explanation is also strong. Deloitte’s CFOs rank cost control (net 62%) above AI (net 47%) as reasons for weaker graduate hiring. The Bank of England’s Agents report limited graduate and entry-level openings against a backdrop of weak demand and elevated labour costs. The US “Canaries” research finds a similar pattern: a 19% employment gap for 22–25-year-olds in highly exposed occupations, working through reduced hiring rather than layoffs. The authors stress the result is descriptive, not causal.

Apprenticeship data show a related shift in who gets trained. In England, starts from August 2025 to April 2026 rose 17.5% for those aged 25 and over but fell 5.4% for under-19s. That fits employers using training budgets to upskill existing staff rather than to bring in new entrants.

Three shapes the ladder could take

The narrow base. Firms hire fewer juniors, rely on experienced hires and AI, and hope to recruit mid-career talent trained elsewhere. This is cheapest in the short run. If every firm does it, the market runs out of mid-career people. It is a classic training free-rider problem.

The compressed ladder. Fewer juniors are hired, but they get real responsibility earlier, directing and checking AI output, handling clients and owning small pieces of work from start to finish. This can produce faster learning, provided firms invest in supervision and feedback. Without that, mistakes spread at AI speed.

The rebuilt rung. Firms deliberately redesign junior work around what AI can’t do: verifying and testing AI output, handling exceptions, gathering facts directly from clients and patients, and building the evaluation sets and knowledge bases the tools rely on. Some firms use AI itself to generate realistic practice cases, as medicine and aviation use simulators.

Which shape dominates is a management choice, and Grant Thornton’s expansion shows firms are not all making the same one.

Implications

For employers, cutting graduate intake saves money this year and creates a shortage of competent mid-level staff later. The firms most likely to regret it are those that need people who can supervise AI on behalf of clients. That supervision depends on judgement built through exactly the work being automated.

For individuals, the safest early-career positions are those with real feedback loops: client contact, regulatory scrutiny, production incidents and work that gets checked. Being able to show you can evaluate and correct AI output, not just produce it, is becoming an entry requirement rather than an advantage.

For policy, the free-rider problem is the core issue. Changes to the Growth and Skills Levy, including restricting Level 7 apprenticeship funding to those aged 16–21 from January 2026, and the Youth Guarantee subsidies affect how cheaply firms can keep a bottom rung. Whether they are enough is untested.

What to watch

  • The ISE Student Recruitment Survey on 14 October 2026, for graduate and apprentice intake by sector.

  • Entry-level hiring by occupation in further DSIT/Cabinet Office and LinkedIn updates.

  • Graduate unemployment in the next HESA Graduate Outcomes release, which last showed 7%, up from 6%.

  • Whether firms that cut juniors start reporting mid-level shortages from about 2028.

Sources

From AI and Jobs: UK, October 2026