With shift work and variable staffing, the task itself is often the same as in an office setting: assessing a malfunction, filling in a schedule, writing a handover, carrying out a check. What changes is the context in which that task takes place. At three o'clock in the morning, there is no colleague to take a quick look. During a peak on Saturday, there is no manager who can approve a decision within five minutes. The same task that can be taken over with oversight on a weekday morning falls back into the category of human work at night, simply because the oversight is not there.
That is the core of this circumstance: shift work and variable staffing do not affect the task, but the third axis of the assessment, namely whether someone is available at that moment to check. A task that scores favourably on three of the eight axes can still remain undone because the fourth axis -- availability of oversight -- is absent during some shifts and not during others.
Some tasks are insensitive to the shift schedule. An initial triage of incoming reports, the preliminary sorting of data, compiling a handover report from sensor data: that work does not require a person to be on standby at that moment, because the outcome can wait until the next assessor is available. Planning work in which AI drafts a schedule based on fixed rules -- leave, collective labour agreement hours, minimum staffing levels -- can generally also be taken over, with a person reviewing the draft at a moment that suits them.
The difference lies in when the assessment needs to take place. If it can be postponed until oversight is available again, shift work changes little about the outcome. If the assessment must happen within the shift itself, then staffing at that moment is decisive.
Tasks that require an immediate response during an understaffed shift shift less readily. A malfunction that immediately requires a decision about shutting down or continuing, a complaint that must be handled on the spot, a safety incident that must be assessed immediately: if there is only one person on the floor at that moment, there is no one to check the AI proposal. That work remains human work, not because AI cannot handle the content, but because oversight is lacking at the moment it is needed.
This also relates to the cost of errors: an incorrect assessment during a night shift with minimal staffing is harder to correct than during the day, and that difference in recoverability carries through into what does AI take over in work with high error costs.
Two companies with identical tasks and identical schedules can still score differently, and the reason is rarely the task itself. The first difference is technical: whether the systems running at night use the same data and the same rules as during the day. If the night shift works with a different system that is not linked to the day system, AI cannot build up a continuous picture and less work shifts, as also shown in what does AI take over if your systems are not linked together.
The second difference is organisational: is there a fixed on-call contact reachable outside office hours to approve or reject an AI proposal? If so, the second category -- oversight for a reason -- remains intact even at night. If that on-call contact is not there, the same task falls back to human work, purely due to staffing.
The third difference lies in the nature of the work itself. Companies where every shift looks different, with many exceptions and specific arrangements per customer or location, see work shift less readily than companies with a fixed pattern, which aligns with what does AI take over if you deliver a lot of customisation. And where the shift involves a lot of direct contact with customers or patients, the assessment carries different weight than for a shift that is mainly process work, as described in what does AI take over if there is a lot of customer contact in the work.
In sectors with a lot of shift work and variable staffing, such as logistics, this pattern is clearly visible: planning and stock control shift sooner than incident handling on the floor during a night shift. An overview of how this plays out per task type is available in what work can AI take over in logistics.
The eight axes on which a task is assessed, including the axis that weighs availability of oversight, are explained on how we assess a task. That page shows that staffing is not a separate category, but one of the factors that shifts the outcome per task.
If you want to know how this plays out for your own schedules and tasks, a first indication can be obtained with the free quickscan: twelve questions, no account needed, resulting in an indication of what share of the hours in your profile can be taken over by AI today. The full work scan, which assesses tasks down to the level of a shift and a time, is still under construction. What happens to work that is freed up remains a decision subject to your own statutory requirements -- this page offers no advice on that, only a description of what shifts and why.
Vraag maar. Ik ken de kennisbank van deze site; wat ik niet weet, zeg ik erbij.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.