AI-assisted versus AI-aware

I've been watching two groups of colleagues over the last year. The first group uses AI constantly - better emails, faster code, summarised meetings, cleaner slide decks. They are more efficient than they've ever been and genuinely happy about it. The second group also uses AI constantly, but pays closer attention to what the tool is doing than to what it's doing for them.

For the first twelve months, the two groups look almost identical from the outside. Same output per week, same quality of deliverables, similar satisfaction. The divergence only shows up when you start asking questions about what their role will look like in two years.

The ergonomic chair problem

There's a version of this pattern from traditional engineering. Between roughly 2008 and 2015, teams invested heavily in tooling - better IDEs, faster build pipelines, richer monitoring. It made individual engineers substantially more productive. It also quietly narrowed the craft. Many of the skills that had been central to a 2005 engineer were no longer load-bearing by 2015. The engineers who learned to use the new tooling got efficient. The engineers who also kept asking why the tooling existed, and what problem class it solved, stayed on top of the craft.

The AI tooling wave is the same pattern, faster and broader. The colleagues using AI mainly to do their current job well are getting paid right now for being efficient at a job that is changing shape in real time. The colleagues paying attention to how the tools are reshaping the work are the ones whose next job is still coherent.

One makes you more efficient. The other keeps you relevant.

Aware is not the same as specialist

A common misreading of this observation is that everyone should retool into an AI specialist. That's not what AI-aware means, and it isn't a practical recommendation for most people.

AI-aware means something narrower and more personal. When you use the tool, you notice what used to be your craft and is now the tool's output. You notice what new craft is emerging in the negative space the tool created. You don't have to quit your job, retrain, or start a startup. You just have to look up occasionally.

Most days I use AI to draft emails, summarise calls, and produce first-pass versions of documents. Some days I notice that writing those things was itself part of how I thought. Those are the days I write the email from scratch, not because efficiency is bad, but because the thinking happens in the friction and the finished artefact is not the point.

A practical test

If you want one piece of career-hygiene advice for the next two years, it is this. For every AI tool you adopt, ask a second question alongside "how does this help me work faster?" - namely, "what part of my current job is this tool quietly taking over?"

If the answer is that it's automating the boring bits, you're fine. Continue. If the answer is that it's increasingly doing the valuable parts of what you used to do, you have some thinking to do - not necessarily about quitting or switching careers, but about where the centre of gravity of your role is moving.

The people using AI to get better at their current role are doing the responsible and obviously correct thing today. They are also, often without noticing, making their role easier to remove. The answer isn't to stop using AI. It's to keep part of your attention on the renovation happening underneath you while you enjoy the new chair.