AI is a work system.
Not just software.

A technically reliable AI system can still create poorly designed work.

AI can redistribute attention, alter decision authority, change work pace, create new monitoring demands, remove familiar cues, increase dependence on alerts, and reshape how workers coordinate with one another. These are classic human factors and ergonomics concerns — now appearing through digital systems.

The assessment therefore needs to move beyond software performance and ask whether the total work system supports safe, healthy and sustainable human performance.

From “human in the loop” to human-centred system design.

Simply keeping a person somewhere in the process does not guarantee meaningful human control. Good design considers what information the person receives, whether the interface communicates uncertainty, how much time they have to act, whether they understand failure modes, and what happens when they disagree with the AI.

This is where cognitive ergonomics, organisational ergonomics, usability and participatory design become central to AI governance.