AI can now produce almost anything in seconds: a first draft, a prototype, a project plan. That is genuinely useful.
It is also quietly changing what good work means, because producing something no longer proves you understand it.
For a long time, output was proof of comprehension. Building was hard, hard meant effort, and effort meant you probably knew what you were doing. AI cut that link cleanly. The making is easy now; judgment is not.
AI hasn't removed the need to understand the user, the business goal, the workflow, the tradeoffs, or the edge cases. And it certainly hasn't figured out what we should choose not to build. That is where the real work is moving.
This is why the future of work is a design problem, and not just for designers. The work that still pays off has moved upstream, to what was always the hardest part: defining the right problem, understanding the system around it, and deciding what good looks like before we hit accelerate. This is why I built the UX workflow in our design system around three lenses for deciding what is worth building, and six stages for building it well.
A sharper problem statement can save weeks of work aimed the wrong way. An hour more with a real user can fix a brief that would otherwise ship wrong. None of that is production. All of it is design, in the older and truer sense of the word: understanding people and systems well enough to decide what should exist.
Designers have lived with this for a while, even when the rest of the organization mostly saw the output. The screens, the documents, the polished deliverables: none of them were ever the value. They were the residue of the thinking. AI can now produce that residue on demand, which is exactly why the thinking underneath it is the one part that hasn't gotten cheaper, for everyone, in every role.
There is a real cost to ignoring this. When we produce faster than we understand, we quietly accumulate work that functions but that no one can fully account for: decisions no one remembers making, systems that run until the day they don't, output that looked finished but was never understood.
Speed without comprehension isn't progress. It's debt we haven't noticed yet.
A simple place to start is to make the thinking travel with the work. Not just the final answer, but the reasoning behind it:
When that reasoning is visible, your understanding becomes reusable instead of trapped in your head. That is what a decision log is for: the tradeoffs written down, so the next team inherits the reasoning instead of rediscovering it. The same is true of everything else that explains a product. Written down, it keeps working after you have moved on.
Understanding before output is the habit I would like us to practice more widely, and it is the thinking behind the design system I lead. I built it to be used, pressure-tested, and reshaped by the teams working in it.
The future of work was never going to be decided by who can produce the most. Production is handled. It will be decided by who understands what they are producing, and who can say clearly what is worth producing at all.
That has always been design work. Now it belongs to everyone.