Capability as infrastructure

Across many discussions about AI, capability is often treated as an individual attribute.

The focus tends to be on what people know, what they can do, and how they can adapt to new tools. Training, upskilling, and reskilling are framed as primary responses to change.

These are important. But they may not be sufficient on their own.

As work becomes more complex, and as AI becomes more embedded in decision processes, capability begins to take on a different character.

It is no longer expressed only through individual performance. It is shaped by the environments in which people operate — the structures, expectations, and systems that support (or constrain) how decisions are made.

In this sense, capability starts to resemble something closer to infrastructure.

Not visible in a single moment, but present in how work is carried out over time.

For example, the ability to exercise good judgment is not only a function of individual skill. It is influenced by whether there is space to question outputs, whether assumptions can be surfaced, and whether reasoning can be examined without penalty.

Similarly, the ability to work effectively with AI is not determined solely by familiarity with tools. It depends on whether processes make room for interpretation, whether responsibility is clearly held, and whether there is a shared understanding of what constitutes a sound decision.

These conditions do not arise automatically.

They are shaped — often implicitly — by organisational design, by cultural norms, and by how success is defined and measured.

If speed is prioritised above all else, different behaviours emerge than if careful reasoning is valued. If outputs are rewarded without attention to process, then the visibility of how decisions are made may gradually diminish.

Over time, these patterns form the environment within which capability is either strengthened or weakened.

This is why capability cannot be understood solely at the level of the individual.

It is also a property of systems.

Seen in this way, developing capability is not only about equipping people, but about shaping the conditions in which they operate — ensuring that judgment, interpretation, and accountability are supported, not eroded, as technologies evolve.

This does not require a complete redesign of existing systems. But it does require attention to how those systems function in practice, and whether they remain aligned with the kinds of decisions they are now expected to support.

As AI continues to evolve, the question may shift from whether individuals are capable, to whether the systems around them are capable of supporting good decisions.

And in that sense, capability becomes something that is not only developed, but also designed, maintained, and sustained over time.