The first AI cost saving is often easy to spot. A research assistant produces a draft faster. A junior consultant creates the first slide pack in minutes. An account executive no longer has to spend half a day preparing for a meeting. A recruiter can source a talent pool before the coffee has gone cold.
What is much harder to see is the capability being lost underneath that efficiency.
Those jobs were never only about the output. They were where people learned the work. Repetition taught them what a good brief looks like, where a dataset starts to fall apart, what a client leaves unsaid, which recommendation will stand up in a room full of sceptics, and when to ask for help.
That is where the next generation of senior judgement has always come from.
The latest evidence suggests the immediate employment story is more complicated than the popular AI narrative. In the New York Fed’s August 2026 survey of businesses in New York and northern New Jersey, 61% of service firms reported using AI in business processes, compared with 40% a year earlier. Yet only 4% of service firms said they had laid people off because of AI in the preceding six months. Fifteen per cent said AI had reduced hiring, while 13% said it had led them to hire more people. More than a third of AI-using service firms were retraining existing workers.
That pattern matters.
The UK presents a related, and potentially more awkward, version of the same challenge. Research from King’s College London analysed applications to 1,090 UK degree programmes from 2020 to 2025 and linked each degree to the AI exposure of its graduates’ jobs using millions of LinkedIn career histories. Since ChatGPT’s arrival, applications to more AI-exposed courses have grown by approximately 6% more than applications to less-exposed courses. At the same time, the researchers found that entry-level labour-market prospects for graduates from those programmes had weakened.
Young people are leaning into the very fields where the first foothold is becoming harder to find. For UK employers, that raises a longer-term talent-supply question. It also puts a premium on building credible entry routes, rather than assuming the market will keep producing experienced people ready for the roles that remain.
CEOs therefore need to look beyond headcount. Has the organisation quietly reduced the number of routes through which someone learns to become excellent?
The succession problem nobody has put on the dashboard
In executive search, the same requirements appear again and again. Clients want senior people with commercial judgement, sector credibility, client presence, resilience, enough technical fluency to challenge an expert, and an ability to turn incomplete evidence into a decision.
None of those qualities appears by magic at director level.
They are built over years of working through imperfect material, seeing senior colleagues handle difficult trade-offs, correcting an analysis that has gone wrong, and learning how the commercial and human consequences of a decision feel in the real world. AI can speed up the work. It can also remove the repetitions that build the instinct.
The HR agenda is beginning to recognise the issue. The Australian HR Institute has made the entry-level role and the future talent pipeline the subject of a September session, specifically naming routine research, analysis, administration, drafting, and coordination as work that has traditionally helped people learn an organisation and develop professional judgement.
The leadership risk will not show up in this quarter’s productivity dashboard. It will appear later, when a business realises it has a thin layer of managers who can run an AI-enabled operation but too few people with the depth to lead clients, build propositions, coach others, or take responsibility when the system gets something wrong.
A hiring market that rewards the wrong signal
There is another danger. When AI makes polished work cheap, employers can become overly impressed by polished work.
A candidate who arrives with a flawless presentation, immaculate research, and instant answers may be highly capable. They may also be unusually good at using tools. The distinction matters. Senior appointments still require judgement under pressure, the ability to expose a bad assumption, and the confidence to tell a client or CEO that the available evidence does not support the preferred answer.
For hiring, the implication is straightforward. Assessment has to move closer to the work itself. Instead of asking candidates only to present a prepared case, give them an ambiguous problem, limited context, competing evidence, and room to show how they think. Ask what they would need to know before acting. Ask what they would not delegate. Ask what they would challenge in an AI-generated conclusion.
The market is moving in that direction. SHRM’s current analysis of talent acquisition argues that AI will pull recruitment away from administration and towards business-facing talent advice, workforce design, assessment, and judgement. It makes the case for measuring “time to capability”, rather than only time to fill.
That is a useful phrase for every executive team. A role is filled when a contract is signed. Capability arrives only when the person, their manager, their tools, and the work around them are set up to produce a result.
Three questions for your next board meeting
- Which early-career tasks are you automating, and which capability did each one used to build?
- Where will future managers learn to handle ambiguity, client challenge, quality control, and commercial judgement if routine work is no longer their classroom?
- Are you hiring for a polished output, or testing for the independent thinking that survives once the tool is taken away?
How well leaders answer those questions will shape the depth of their future leadership bench. The short-term efficiency gain is obvious. The harder question is whether organisations are still building the experience they will need a few years from now.
This piece first appeared in Talent Pools Nexus, our weekly read for leaders in research, insights and consulting. Subscribe on LinkedIn.
