What the AI Revolution Really Asks of Knowledge Workers

There is a quiet panic running underneath every conversation about AI and work. People sense that something is being taken from them, but they can't quite name it. So they argue about job losses, about which roles are "safe", about whether the machines are truly intelligent or just convincing. These are the wrong questions. The right one is simpler and more demanding: what is the work that only you can do, and are you climbing toward it?

The prank

Here is the prank AI is playing, and it is a cruel one. Everyone assumes the machines will make intellectual work easier — that you'll type a prompt, collect the output, and coast. The opposite is true.

For years, a great deal of "knowledge work" was quietly mimicry: going through the motions, reformatting the known, looking busy, producing the expected artifact without much real thought behind it. That was the comfort zone, and plenty of careers lived comfortably inside it. AI eats exactly that work first.

What's left for you is only the hard part: the actual thinking. So AI does not lower the difficulty of your work; it raises it. It kicks you out of the comfort zone, strips away every task you could once do on autopilot, and leaves you face to face with the work that forces you to evolve. The tool that was supposed to relax you is the one that won't let you hide.

The floor is rising

Start with what is actually happening. AI does not eliminate knowledge work in one stroke — it commoditizes it from the bottom up. The lower-level outputs of any knowledge profession — the first-draft summary, the boilerplate analysis, the routine memo, the standard recommendation, the data pulled and tidied — are becoming abundant and nearly free. Whatever could be produced by following a known procedure is now the cheapest thing in the building.

This is not a one-time event. It is a rising floor. Each year the waterline of "things a LLM can do well enough" moves up another level. Work that felt skilled and defensible in 2022 is table stakes in 2026. The commoditization is continuous, and so the only stable response is also continuous: keep climbing to the next level of abstraction before the floor reaches the rung you're standing on.

Every level has to move up

Here is the part most people miss. This is not just a problem for junior staff. It applies to all levels. The analyst whose value was assembling the model, the manager whose value was synthesizing the analysts, the strategist whose value was packaging the synthesis — every one of them is being asked to partially or even completely vacate their current rung, because AI-assisted workflows, more or less autonomous, are moving in beneath them to fill it.

You don't get to opt out by being senior. Seniority just means you have further to climb and more habits to unlearn. The direction of travel is the same for everyone: away from routine execution and toward abstractive and relational thinking — the ability to see the shape of a problem no one has framed yet, to connect domains that don't obviously touch, to hold ambiguity and judgment where there is no procedure to follow.

Where AI stops being useful

And this is the genuinely good news, the thing worth holding onto. The higher you go into abstractive and relational thinking, the less useful AI becomes — and the more your own raw human intelligence matters.

AI is extraordinary at the levels with structure: known patterns, retrievable knowledge, well-posed tasks. But at the top of the ladder — original framing, taste, the leap that reframes a question, the judgment that weighs incommensurable things — the models thin out fast. That territory still belongs to human intelligence, and it will for a long time. The ladder isn't a trap closing on you. It's an invitation. The summit is the safest place to stand, and it happens to be the most rewarding place to work.

The strategy: climb, and protect your attention

So what do you actually do, whoever you are, whatever you do? The strategy is the same for everyone.

Climb as far as you can. Treat your own intelligence like an athlete treats their body — something to be coached, trained, and pushed deliberately, not left to drift. Compounding personal growth is the only asset that keeps appreciating while everything below it is commoditized.

Prioritize ruthlessly. You cannot climb while drowning in low-level tasks. Use AI to declutter — to absorb the routine, the repetitive, the procedural — so your scarce hours go only to the directions that compound. Invest your effort where it yields growth on growth; let everything else fall to a pipeline.

Delegate the rest to AI. Anything that can be turned into a workflow should be. Treat AI not as a magic answer box but as infrastructure — a set of pipelines you build to declutter your attention going further.

And here is the move that ties it together. Building those pipelines is itself the first high-yield investment. The hours you spend designing the systems that free you are not overhead — they are the highest-leverage thing on your calendar, because they convert your future low-level work into something automatic and hand your attention back to you.

Which brings us to the real point. Beneath money, beneath time, beneath skill, the most important asset a human owns is attention. It is finite, it is non-renewable within a day, and it is the only input that can be aimed at the top of the ladder where machines can't follow. The AI revolution is, in the end, a forcing function: it strips away every excuse for spending attention on the commoditizable, and rewards — lavishly — anyone disciplined enough to point it at what compounds.

Climb as high as you can. Guard your attention like the asset it is. Let the machines have the floor.