Kennisbank

AI and management: what changes in the tasks?

Management may at first glance seem like a domain of consultation, judgment and responsibility — all typically human territory. Yet a management function consists of a series of separate tasks, and these tasks differ greatly in the extent to which AI can take them over. Of the 40 tasks we distinguish within this domain, a clear portion falls into each of the three categories. The pattern behind this is more useful than a checklist.

Where AI can already take over the work

Tasks that mainly consist of collecting, organising and summarising existing data lend themselves well to full takeover by AI. Think of drawing up a management report for the board or shareholders: the figures are already in the systems, the structure of the report is largely fixed, and AI can flag deviations and trends without a human having to type out every line. Monitoring budgets and flagging deviations also falls into this category: it is a matter of comparing actual expenditure with the budget and applying threshold values, something software is structurally better at than a human who does it alongside other work on a weekly basis.

The common feature of this category is that the outcome can largely be determined objectively from available data, and that the task repeats according to a fixed structure. Little room for interpretation is needed and no weighing of interests. That makes the difference with comparable tasks in other departments not coincidental: at can AI take over keeping an audit log of financial changes we also see that systematic, data-driven registration is readily automatable, while interpreting what that registration means requires something different.

Where AI supports, but oversight remains necessary

A large part of management work sits between full automation and fully human work. Here AI can produce a concept, analysis or proposal, but a human assesses the result and approves or rejects it, for a reason.

Drawing up an annual plan with concrete goals is a good example. Based on historical data and input from the organisation, AI can create an initial draft of goals, actions and responsibilities per department. But whether those goals are realistic, ambitious enough and mutually consistent requires assessment by someone who knows the organisation, its people and its constraints. The same applies to setting KPIs per team or process: AI can make suggestions based on comparable organisations or industries, but the choice of which indicators should truly be steering depends on priorities that cannot be fully derived from data.

Preparing and setting the agenda for management meetings also falls into this middle category: AI can gather documents, propose an agenda and produce summaries, but someone must judge whether the right topics have been placed on the agenda in the right way, and whether sensitive points should or should not be discussed in this particular setting. The pattern is always the same: AI delivers a solid first version, a human tests it against context that is not fully captured in data, and approves or rejects it.

Where the work remains human

A third group of tasks largely remains with people, not because AI cannot process the information, but because the task itself revolves around taking responsibility, weighing interests or speaking on behalf of the organisation.

Making decisions during a management meeting is the clearest example here. Fully analysed information may be on the table, but the choice itself — with the responsibility that comes with it — lies with the people who have the authority and the mandate. AI can make options and consequences clear; someone else presses the button.

Drawing up a strategic plan for the coming years falls largely into the same category. AI can supply market, customer and internal analyses, but choosing direction for an organisation over the medium term requires weighing risk, ambition and what the organisation actually wants to become — something that cannot be objectively derived from data. Planning staffing needs at management level is sensitive for a different reason: it directly touches decisions about people and roles. Such decisions are subject to their own statutory requirements that an employer must comply with, and these require a diligence that cannot be replaced by a task analysis. Those seeking a broader view of staffing issues will find a similar distinction on the page about what work can AI take over in the HR department: there too, the line between supporting data analysis and decisions about people can be drawn sharply.

The pattern behind the classification

Looking at the 40 tasks in this domain, a line can be drawn: the more a task consists of the repeatable processing of existing, structured information, the sooner AI can take it over fully. The more a task requires combining data with organisational knowledge that is not documented, the sooner oversight is needed. And the more a task itself constitutes the exercise of responsibility, mandate or judgment — not the preparation of it — the sooner it remains human work. This pattern is not unique to management: the same logic explains, for example, why at what work can AI take over in logistics planning tasks are more quickly automatable than tasks where someone has to agree an exception with a supplier or customer.

This classification says something about the tasks, not about how much management an organisation needs or how it structures its staff — that is and remains up to the organisation itself, within its own applicable statutory frameworks.

What you can do now

Would you like to know how this classification translates to your own situation? The free quickscan from ftetoai consists of twelve questions, works without an account, and gives an indication of which portion of the hours in your profile can be taken over by AI today. Please note: this is an initial indication, not a full analysis. The extensive work scan that goes deeper into task level is still under construction — we cannot offer it to you yet, but the quickscan already provides a fair starting point.