Most of the anxiety about AI in knowledge work is aimed at the wrong thing.
Every profession has constants and variables. The variables are the tools, methods, and workflows required to execute at any given moment. The constants are harder to name but easier to recognise: the instincts, the judgment, the accumulated understanding that made you effective before you learned the tool you're using now.
The anxiety is almost entirely about the variables.
Designers worry that AI will produce better UI than they can. Writers worry about drafts being generated. Developers worry about code being auto-completed. These are concerns about the current medium of the work. Not the work itself.
And UI is one part of what a designer does. So is UX. The judgment underneath both is different. Knowing what to build, for whom, and what the experience should feel like. AI doesn't replicate that.
What AI actually creates is speed. And speed creates affordance. A designer who can generate UI variations in seconds isn't being replaced. They're freed from the part of the job that used to eat most of the time. That time can go somewhere better. The micro-interactions that usually got cut. Personalisation that wasn't possible at scale. The small details that separate a product people use from one they prefer. This work always mattered. There just wasn't time for it.
Some UI work will be produced by AI directly. Some workflows will disappear entirely. That has happened before and it will happen again. The people who navigated previous shifts were not the ones who held on to the old tool. They were the ones whose ability existed independent of it.
The constants never lived in any tool. Understanding the people you're solving for. Holding the tension between what a business needs and what a person actually experiences. Recognising a problem nobody has named yet. These accumulated through years of paying attention, and they transfer.
Fluency with a tool is easy to mistake for professional ability. They are related but not the same. Someone fast in Figma is not necessarily a good designer. Someone who writes clean code is not necessarily solving the right problem. The tool-level skill is a variable. The instinct underneath it is a constant.
The real risk is not that your tools become obsolete. It is defining yourself by the variables of your profession. If your professional identity is "I'm a UI designer" or "I'm a React developer," the identity breaks every time the tools or methods change. If it is closer to "I understand how people navigate complex decisions" or "I know how to find the gap between what users say and what they need," you are standing on something more durable.
AI is a variable. Treat it like one. Learn it, use it, and spend the time it frees on the things that always got cut: the details, the personalisation, the work that was too slow to justify before.