What changes when AI becomes normal at work?

A few things I’ve written recently keep turning out to be the same question from different angles. This is the working thread that connects them, not a new argument.

I’ve started to notice that a lot of what I’ve been writing lately might be parts of the same question: what happens to knowledge work once AI stops being something you experiment with and becomes part of how you work every day?

A few threads keep recurring.

Job descriptions start to feel too narrow once one person can move easily across research, writing, analysis, strategy and building — which changes what a manager is for more than it removes the need for one.

Prompting starts to matter less than routing: knowing which model, tool, source of context or person a piece of work should actually go to.

If AI can generate almost unlimited options, filtering matters more than generating. Capability is becoming abundant enough that context and judgement are the scarce things, not raw model intelligence.

And then there’s the apprenticeship problem. If AI absorbs the routine work people used to learn through, where does experience actually come from?

I don’t have a thesis connecting these yet. But I’m starting to think a lot of the AI conversation is still focused on tools and productivity, when the more interesting question is what happens to the structure of work around them — job descriptions, management, apprenticeship, attention.

So I’m keeping this as a working thread rather than forcing it into an argument before it’s ready.

AI may be changing the unit of work faster than organisations are changing the systems around it.