Corporate Training and AI: Who Will Train the Managers?

The 78% that describes a system failure

Talogy research published on 22 September 2026 gathers responses from more than 200 senior leaders in HR, talent acquisition and learning and development. Some 78% report concern about the development of critical leadership skills as entry-level tasks shift to artificial intelligence systems.

The sample is what makes the figure interesting. Respondents govern hiring, internal career paths and succession plans: they observe the mechanism from inside the machine that produces it.

The wording of the research remains precise: important managerial skills "risk developing less naturally than in the past", because the number of people entering the leadership pipeline is shrinking. The summary published by HR Dive^[1] records the same concern across every function surveyed.

How organisations used to produce their own managers

For decades, organisations produced their own managers as a by-product of ordinary work.

The path was recognisable. Someone joined, received narrow and repetitive tasks, made mistakes on low-risk material, and learned to read a process from the inside. Three or four years later they were coordinating other people on that same process.

Formal training came afterwards and certified a competence already acquired in the field. The real learning sat in the boring tasks: the review, the reconciliation, the first draft rewritten five times.

Those tasks are precisely the first to be handed over to automated systems today. The delegation has sound economic logic within the current quarter. The cost shows up elsewhere, five or seven years later, when a fully formed middle management layer is needed.

The skills that mature through practice

Talogy also worked with a larger sample: more than 6,000 people, managers and talent professionals across several countries. Ali Shalfrooshan, the company's director of science strategy, identifies the capabilities that sustain a leadership pipeline over time.

• learning mindset
• critical thinking
• adaptability and resilience
• clear communication
• collaborative thinking

The list looks sober, almost obvious. The difficulty lies in how these capabilities mature: a course introduces them, repeated practice embeds them.

The research insists on one operational point. Organisations must intentionally replace the learning opportunities that daily work once offered spontaneously. The adjective carries weight: replacing requires design, budget and someone accountable for the outcome.

The official statement issued by Talogy^[2] puts it bluntly: a more intentional approach is needed in place of natural on-the-job learning.

The entry door and the floor above

The contraction at the entry level is already visible in labour market data. Federal Reserve Bank of New York figures have shown for months an unemployment rate among US recent graduates higher than the general average, a rare inversion.

This evidence is almost always discussed as a question of generational fairness. The Talogy research moves the argument up one floor.

The squeeze on entry-level roles produces two distinct effects. The first hits those looking for a first job today. The second hits the organisation five years from now, when coordination positions will call for people who grew up in those entry-level roles.

The second effect arrives quietly and surfaces as a sudden shortage of credible internal candidates. The same research, picked up by the Manila Times^[3], is read as a global signal about leadership pipelines.

The objection: AI accelerates learning

There is a reasonable objection. Generative systems can compress learning times: they explain a process, simulate a scenario, give immediate feedback on a decision.

The objection holds technically and fails organisationally. The skills identified by the research (judgement, adaptability, the ability to hold together people who disagree) mature through exposure to real consequences.

A simulation teaches the procedure. Accountability teaches the craft.

A second limitation concerns latency. Whoever decides today to automate the entry level will see the effect on leadership once the current executive group has moved on, and the correlation will be hard to reconstruct after the fact.

This time lag explains why the concerned 78% coexists with flat development budgets. The research therefore calls for an explicit choice about the perimeter of automation.

What organisations that design the entry level are doing

Some organisations are responding, and their method shares common traits.

They rebuild learning opportunities as design objects, with the same care given to a production process. They assign junior people tasks that systems could perform, when those tasks produce competence.

The choice costs something in immediate efficiency. It is treated as an investment in human capital, with a stated horizon.

The second trait concerns internal conversion. Training and redeploying people already on the payroll proves cheaper than the layoff-rehire cycle, and it builds exactly the continuity of experience that the leadership pipeline requires.

The third trait is measurement. These organisations count how many coordination positions are filled from within, and read that number as a health indicator.

A fourth trait concerns middle managers. They receive protected time to teach, and that time appears in their evaluation as an outcome, alongside the period's numbers.

The reading for decision-makers

The evidence changes the conversation at the four tables where decisions are made.

• CEO: the organisation's capacity to generate its own leaders
• CHRO: the design of entry-level roles
• CFO: the comparative cost of internal development versus external search
• Talent and remuneration committee: the share of internal promotions

For the CEO, the question to the board concerns the organisation's capacity to produce its own leaders over the next five years. For the CHRO, the priority becomes the design of entry-level roles, ahead of the training catalogue.

The CFO finds here a line item with a documentable return. The cost of an internal plan can be set against the full cost of an external search for middle management profiles.

The talent and remuneration committee has the simplest metric: the share of internal promotions among managerial positions, read year on year.

These four readings share one premise. The leadership pipeline is infrastructure, and every piece of infrastructure degrades quietly when maintenance stops.

The design question for the next twelve months

The 78% recorded in September 2026 is a figure to take into the boardroom, rather than a press-clipping headline. It describes a group of professionals who can see a shortage coming and are working with tools designed for another era.

The distance between the concern expressed and the budget allocated is the space where this problem lives.

The design question admits a verifiable answer: which tasks does your organisation deliberately assign to junior people, for the purpose of producing the leaders of 2032?

A documented answer is worth more than an announced training programme. The rest is a bet that managerial competence will generate itself, inside an organisation that has just removed the occasions for generating it.