Why AI Is Not Killing Jobs. It Is Killing the Path to Knowing How to Do Them.

Audience and Use Primary reader: CEO, CHRO, Board member, Transformation lead, Head of Talent Primary use case: Board pack insert, Workforce strategy offsite pre-read, Executive briefing on talent pipeline risk

In the last twelve months, your organisation probably did three things. It approved a generative AI budget. It launched at least one pilot. And it told the board that headcount would hold steady because AI was making every team more productive. The quarterly numbers backed you up. Output per employee climbed. Delivery timelines shortened. Graduate hires were cut or frozen, because the remaining team, augmented by AI, seemed to be doing just fine.

Nobody raised a concern. Nobody asked what those graduates were supposed to be learning. Nobody noticed that the work which used to turn a 23-year-old into a capable 30-year-old had quietly disappeared from the workflow.

"The common view is that AI threatens jobs. That is the wrong target. AI is removing the developmental experiences that make people ready for the jobs that remain."

If organisations keep optimising for output without protecting for learning, they will hollow out their leadership pipelines in three to five years and find themselves unable to replace the senior people they depend on today. This is not a workforce planning footnote. It is a structural risk to organisational capability.

Drift vs Design

Drift is what happens when AI adoption follows efficiency logic alone. Nobody decides to destroy the talent pipeline. It erodes by default, one automated task at a time, one frozen graduate role at a time, one quarter of rising output at a time.

Design is what happens when leaders treat human development as a system constraint, not a nice-to-have. It means deliberately protecting learning friction even when AI could smooth it away. It means measuring capability growth alongside productivity. It means asking, before every automation decision, not just "Can AI do this?" but "What does a human lose by not doing it?"

Definition: The Missing Rungs Problem

Term: The Missing Rungs Problem

Definition: The systematic disappearance of entry-level and early-career work experiences through which employees historically developed the skills, judgment, and resilience needed to advance into leadership roles. Caused by AI and automation absorbing the formative tasks that once served as the training ground for professional growth.

Why it matters: Without these developmental experiences, organisations produce employees who can deliver AI-augmented output today but cannot independently lead, decide, or problem-solve tomorrow. The result is a leadership pipeline that looks full on paper but is structurally empty.

How it shows up: Declining internal promotion rates. Rising external hiring costs for mid-level and senior roles. Junior employees who hit a capability ceiling after two to three years. Growing dependency on a small number of senior staff who hold irreplaceable institutional knowledge.

The opposite: Deliberately designed early-career pathways where AI augments learning rather than replacing it, and where organisations measure capability development alongside productivity.

How the Rungs Vanish

AI is accelerating the automation of entry-level tasks, but what is often missed is the type of work being automated. It is frequently the grunt work and glue work that juniors historically handled as they learned their craft. Junior analysts once spent hours preparing reports, checking data, drafting documents, answering routine queries. Now AI tools can produce a polished report or code snippet in seconds. On the surface, this seems purely beneficial for efficiency. But those tasks were also training exercises. By taking them over, AI is compressing entry-level work. The work still gets done. The learning opportunity disappears.

AI does not just remove drudgery. It removes the developmental struggle. Learning happens through trial and error: a junior employee attempts a task, receives feedback, corrects mistakes, gains competence. AI now often provides the correct answer or automates the process entirely, skipping the struggle. The invisible learning loop is cut short. What is being automated away is the practice ground for expertise.

A recent Stanford study of US payroll data found that entry-level hiring in AI-exposed jobs has sharply declined since late 2022. Young workers aged 22 to 25 in highly AI-exposed roles, such as software development, customer support, and junior analyst positions, saw employment fall 13 per cent in just a couple of years, while older workers in the same roles were largely unaffected. In less AI-exposed occupations, young worker employment actually rose. Companies are not eliminating jobs across the board. They are stripping away the junior roles in functions susceptible to AI.

From the employer's perspective, an entry-level worker boosted by AI might appear remarkably productive. With generative AI, a single junior employee can generate content or code with the apparent proficiency of a much more experienced person. But this is a productivity mirage. The outputs look good. The junior is not developing the underlying skill to produce that work independently. AI can create the appearance of 10x workers while removing the 10x learning. Individual tasks get done faster, but the individual does not gain the expertise that doing those tasks without AI would have provided.

Many organisations are unknowingly encouraging this. Over 80 per cent of firms have staff using AI tools informally, self-reporting 20 to 40 per cent productivity gains. This shadow usage can make each employee more efficient, but it often happens in isolation. People are not asking each other for help as often when an AI assistant is available. The shared struggles that bonded junior and senior coworkers are disappearing, further thinning out learning opportunities. Organisations become output-efficient in the short term, but skill-deficient in the long term.

Leadership Capability Is Being Hollowed Out

If the foundational rungs of a ladder disappear, the top eventually has nothing supporting it. Today's senior managers and experts gained their skills by climbing those very rungs that are now missing. They had years of seasoning through progressively challenging assignments. If new employees are not getting comparable experiences, who will be ready to step into leadership roles five or ten years from now?

In the Stanford analysis, employment of senior workers aged 35 and above in AI-heavy jobs actually rose 8 per cent since 2022 even as junior roles declined. Companies are retaining experienced staff who hold critical knowledge but scaling back junior hiring. It may seem fine for now. Projects still get delivered. But it masks a looming succession problem. No junior roles means no pipeline. No pipeline means today's seniors become tomorrow's bottleneck.

Many organisations will not realise the severity until it is too late. The Missing Rungs Problem does not show up in quarterly metrics. In the near term, everything looks normal or even more productive. Junior employees still turn in work, augmented by AI, and short-term output metrics may improve. But the qualitative depth of their experience is poorer. By the time companies notice, they have a leadership bench that was never built. Succession plans become perilous, forcing firms to look outside for talent to fill mid-level and senior roles.

The warning signs are already visible. Internal promotion rates are declining. Critical positions stay vacant longer, prompting expensive external searches. Companies are flattening hierarchies or eliminating middle-management roles under the banner of efficiency. In 2023, Accenture cut 19,000 jobs even as it hired new AI specialists. IBM eliminated nearly 4,000 jobs and announced plans to replace thousands more with AI. McKinsey downsized while doubling its AI consulting workforce. Traditional roles out, AI-centric roles in, at a fraction of the headcount. Fewer total employees may boost profits temporarily, but it also means fewer future leaders in the making.

A survey of employers found that 70 per cent believed AI could already replace an intern. Entry-level job postings fell by about 35 per cent since early 2023 in AI-affected sectors, and internship openings in the UK dropped roughly 30 per cent since 2022. Fewer openings for newcomers mean fewer future managers. Internships, analyst programmes, and apprenticeships, the traditional learning-by-doing pathways, are evaporating.

Over time, organisations face a scenario of professions without apprentices. A law firm with plenty of seasoned partners but no associates being trained underneath them. A hospital with senior surgeons but a shortage of residents coming up the ranks. Expertise cannot reproduce itself if there is no apprenticeship. The profession hollows out from the bottom up. The much-maligned middle manager was often seen as expendable, but they play a key role in transmitting institutional knowledge and culture. A company that automates or eliminates many mid-level roles loses that glue.

If companies cannot promote from within, they face higher costs to recruit outsiders who have the requisite experience. External hires often command higher salaries and take time to onboard. They do not carry the same institutional memory. A company that has not built its bench will have to buy its talent later, an expensive and risky strategy. This cost compounds over time. The longer a company ignores the development gap, the more it will pay down the road. What looks like efficient automation today can seed a leadership crisis for tomorrow.

A Structural Challenge, Not a Generational Failing

The Missing Rungs Problem is not about flaws in the new generation of workers. This is a structural challenge driven by how technology is reconfiguring work itself. Young people want to grow. They simply are not given the same ladder to climb. The Stanford researchers pointed out that the data does not show age-based bias. The roles that young workers would normally step into are the ones being automated or eliminated first. The result can look like generational disparity, with Gen Z struggling to launch careers, but its root cause is structural and technological, not attitudinal.

When entry-level roles vanish, even the most talented and well-educated young candidates struggle, no matter how motivated they are. We are already seeing a 50 per cent drop in hiring of new college graduates in some sectors since 2019. With advanced AI, the effect is accelerating. A cohort of would-be future leaders is underemployed or rerouted into jobs unrelated to their training. It is not because Gen Z lacks work ethic or skills. Entry-level jobs have changed: they now demand higher starting skills, involve managing AI outputs, or simply are fewer in number.

Why Training Cannot Fix This Alone

Faced with a talent pipeline problem, a typical corporate response is: "We will just train our employees more. We will upskill the juniors we have." Training alone is not a sufficient solution. The reason lies in the distinction between formal training and experiential learning. Traditional training programmes, workshops, online courses, bootcamps, are helpful for knowledge transfer, but they cannot replicate the learning gained through real work experience over time. A huge portion of professional development comes from on-the-job experiences: learning by doing, stretch assignments, mentorship. If those on-the-job opportunities dwindle, no amount of classroom instruction can fill the gap.

Consider what is lost when a junior employee no longer writes first drafts because an AI does it. You could put that employee in a writing course, but it is not the same as writing, getting edited, and iterating in a live job context. The muscle memory of problem-solving, client communication, and project management develops in real situations, dealing with a tricky client email, fixing a mistake under deadline, brainstorming from scratch. If AI shields juniors from ever experiencing those challenges, a two-day seminar will not confer equivalent resilience or insight.

You cannot train someone on a job that no longer exists, and you cannot fully replace on-the-job learning with classroom training. The ladder itself needs rebuilding.

If half of entry-level roles disappear, you cannot retrain half of fresh graduates into senior roles. Experience and seasoning do not work that way. Training and mentorship remain vital, but they must be complemented by deliberate changes in work design. Companies will need to create intentional practice arenas for junior talent, rather than assume existing job structures will naturally develop them.

Wrong Approach vs Right Approach

Wrong: Measure AI success by output volume alone. Right: Measure AI success by output and capability growth together.

Wrong: Freeze graduate hiring because AI covers the work. Right: Redesign graduate roles around judgment, communication, and decision-making that AI cannot do.

Wrong: Let junior employees use AI without structured learning requirements. Right: Require junior employees to demonstrate independent competence before and alongside AI use.

Wrong: Treat internal promotion decline as a market problem. Right: Treat internal promotion decline as a pipeline design failure.

Wrong: Assume training courses can replace on-the-job experience. Right: Design deliberate practice arenas inside the workflow where juniors learn through real stakes.

Wrong: Flatten hierarchy to cut cost. Right: Protect the middle layer that transmits knowledge and mentors the next generation.

The Cost of Getting This Wrong

• Succession debt. Every year without a functioning leadership pipeline adds twelve to eighteen months to future replacement timelines. This debt compounds.

• External hiring premium. When you cannot promote from within, you pay thirty to fifty per cent more to recruit externally for mid-level and senior roles, plus onboarding time and cultural risk.

• Institutional memory loss. Senior staff who leave take with them knowledge that was never transferred, because the junior staff who would have absorbed it through shared work were never given the chance.

• Productivity mirage collapse. The short-term output gains from AI-augmented juniors mask a growing gap between what they deliver and what they understand. When they are asked to lead, the gap becomes visible.

• Organisational brittleness. A company with a thin leadership bench cannot absorb shocks. One resignation, one restructure, one market shift exposes the gap.

Pipeline Health Diagnostic

Answer these five questions honestly. Score one point for each "yes."

1. Has your organisation reduced graduate or entry-level hiring in the last eighteen months while increasing AI tool deployment?

2. Can you name specific tasks that junior employees used to do for learning purposes that are now handled by AI?

3. Has your internal promotion rate for mid-level roles declined or stalled in the past two years?

4. Are more than forty per cent of your critical senior roles filled by people within five years of retirement or likely departure?

5. Does your organisation measure productivity per employee but not capability growth per employee?

Score 0 to 1 (Green): Pipeline appears healthy. Monitor annually. Ensure AI deployments include a learning impact assessment.

Score 2 to 3 (Amber): Early erosion is likely. Audit your last three AI deployments for learning displacement. Redesign one graduate pathway this quarter.

Score 4 to 5 (Red): Pipeline starvation is underway. Commission a succession risk review within 30 days. Begin designing deliberate practice arenas for junior staff immediately.

If you are in Red, do this next: convene your CHRO and two most senior operational leaders. Map every role that currently depends on one person. That is your exposure.

The Evidence Base

What we know

• Entry-level hiring in AI-exposed occupations has declined sharply. Stanford Digital Economy Lab data shows employment of 22- to 25-year-olds in highly AI-exposed roles fell 13 per cent between late 2022 and 2024, while older workers in the same roles were largely unaffected.

• Entry-level job postings have dropped roughly 35 per cent in AI-affected sectors since early 2023. UK internship openings fell approximately 30 per cent since 2022.

• Seventy per cent of employers in a recent survey said AI could already replace an intern.

• Over 80 per cent of firms report employees using AI tools informally, with self-reported productivity gains of 20 to 40 per cent.

What we do not know yet

• How long the lag is between learning compression and visible capability gaps. Early indicators suggest three to five years, but this will vary by sector and role complexity.

• Whether deliberately designed AI-augmented learning pathways can fully compensate for the loss of traditional experiential learning. The evidence is too early.

• The differential impact across sectors. Knowledge work is clearly affected. The picture in trades, healthcare, and public sector is less certain.

What I have observed in organisations

• The firms that adopted AI fastest are often the ones with the weakest awareness of learning displacement. Speed of deployment and depth of human impact assessment are inversely correlated.

• Junior employees are rarely aware of what they are not learning. The gap is invisible to the person experiencing it.

• Middle managers are the earliest detection system. When they report that junior staff "can't think independently" or "always need the AI," this is a pipeline signal, not a performance complaint.

The Strongest Objection

The most credible pushback is this: "AI is a tool, and every generation of tools has changed how people learn. Accountants no longer learn with ledger books. Designers no longer learn with T-squares. The learning adapts." This is fair. It is also historically accurate. What makes this moment different is speed and breadth. Previous tool transitions took a decade or more, giving professions time to evolve their apprenticeship models. AI is compressing that transition into two to three years, across nearly every knowledge-work function simultaneously. The risk is not that learning will never adapt. It is that adaptation takes time, and organisations are not buying themselves that time. They are simply absorbing tasks and assuming development will sort itself out. History says it will. History also says the lag can be a generation long, and many organisations will not survive the gap.

"AI is not removing jobs. It is removing the instructions for how to get one."

"You cannot train someone on a job that no longer exists, and you cannot replace on-the-job learning with a classroom course."

"No junior roles means no pipeline. No pipeline means today's seniors become tomorrow's bottleneck."

"The goal is not more technology. It is more capable humans. In a world where AI is ubiquitous, investing in human capability is the only sustainable competitive advantage."

The Question that Remains

Every organisation will eventually face this reckoning. The ones that face it now will design their way through it. The ones that wait will discover that the people they need to lead the next chapter were never given the chance to learn how.

The rungs are disappearing. The question is not whether your organisation is affected. It is whether anyone in your leadership team is looking at the ladder.