AI: Management Is Dead – Long Live Management! — An Opinion Piece by Anne-Laure Pams for FocusRH
21.09.2026
Can a machine fire someone? Three years ago, the question might have raised a smile. Today, no one is laughing. Artificial intelligence is no longer merely assisting managers: it is encroaching on their most sensitive territory — authority. Meta has been accused of collecting employee performance data with a view to preparing redundancies. At one Bay Area company, an AI system issued disciplinary warnings to an employee before recommending their dismissal. These are no longer hypothetical future scenarios; they are real events. And they raise a question that few are prepared to ask outright: do managers still serve a purpose?
Legitimacy Under Threat
Managers traditionally derived their authority from three pillars: they knew, they controlled and they made the decisions. All three are now being shaken at once.
They knew more than their teams — or were expected to. That position as the person “in the know” collapses the moment an employee can access greater expertise than their manager’s at the click of a button.
They controlled the work: reporting, deadlines, task allocation and productivity measurement. AI can now automate a large proportion of these activities, at a scale and speed never seen before.
They made the decisions. Yet the two examples above show that some organisations are already delegating disciplinary judgement to algorithms.
We should be clear-eyed about this: if management amounts to no more than that, it no longer has any reason to exist. And the temptation to pursue short-term savings is very real: replacing an expensive layer of management with a system that is available around the clock is a calculation that is already being made in some executive committees.
Leadership: The Only Real Added Value
What remains is precisely what AI cannot do: set a direction, bring a team together around a shared sense of purpose, build trust, make judgements amid uncertainty and, above all, take responsibility for the decisions made. That foundation cannot be delegated to a machine.
It has never been rarer — and therefore never more valuable.
Recent examples of errors produced or amplified by AI illustrate the limits of delegation without human oversight.
In the case of the Californian company, the system had failed to apply correctly a rule that it had itself helped to formalise.
In New Zealand, a report bearing Deloitte’s name is said to have relied on outdated data, with a potential error estimated at $21.5 million. These two failures do not mean that “AI is incompetent”; rather, they show that “AI without human oversight can become dangerous”.
An analysis can be sophisticated, data-rich and highly persuasive in its presentation — yet fundamentally wrong. Checking source data, questioning assumptions and challenging conclusions: this is work that no machine can delegate to itself.
What appeared to weaken the manager’s position in fact makes their role more legitimate: without human critical judgement, AI amplifies its errors rather than correcting them.
The manager of tomorrow will look less like a traditional boss and more like an airline pilot: even if the machine handles most of the flight, they must be ready to take back control at the first sign of turbulence. Automation does not eliminate their role; it shifts their expertise away from execution and towards judgement and accountability.