It's hard to move in the regular spaces right now as a designer without noticing the continuing noise about AI. It's loud at best, and braggadocios at worst.
Six months ago I was anxious. Reading LinkedIn posts at midnight wondering if my career was about to get automated out from under me. I’ve been designing for a long time, and suddenly the space felt unfamiliar.
I felt pressured to take a position on AI before things have settled, while smarter people than me at all levels litigate the details. I was worried about the pace of change and seeing every day lost as a compounding pile of work to do to catch back up
Well, I’ve spent the last month trying to catch up and I have news. Things haven't settled, the ground is still moving, and sitting with uncertainty might be the best position right now.
You control how much AI enters your work
AI is a big concept and the signal-to-noise ratio is terrible, but cut through the noise and boil it down – it's a big collection of tools under a single umbrella, and when it comes to adopting any new tool, you are the gatekeeper.
You wouldn't adopt a new approach or tool without being sure it had a positive outcome for your work. AI is no different, and it doesn't get a free pass just because the discourse is louder.
For me, any stance on a new tool in my work is valid if I can say the work I’m doing is better with it than without it, and, at the barest minimum, the quality of the outcomes for users is unaffected.
And the opposite is true. If a tool doesn't clear that bar, I won’t use it.
You've been evaluating tools your whole career. This is the same skill applied to a noisier landscape.
Wait six months
I've got a theory that AI hype can be countered with patience. The day I hear the hype, I don't worry. If it's worth knowing it will be around in six months.
In six months the early adopters will have learnt all the lessons, the analysis will be done, and the answer to "should I pay attention to this?” will be clear. Six months is no guarantee it will be appropriate for me, but making it past the patience filter is a strong signal.
For example, I first heard about Claude Code in early summer 2025. There was a lot of noise, the web was thick with think pieces and reaction posts, and it looked interesting.
So I ignored it.
Six months later, in the run-up to the Christmas break, I took notice again, because people were still talking about Claude Code. People I knew in real life were talking about Claude Code. It was time for a look.
The think pieces and bluster were gone. What was waiting for me was established tutorials and shared lessons, and best of all, a fresh new model that was even better than when it first launched.
And the thousands of other tools that landed with great fanfare, only to disappear? Gone, without me having spent a minute of my time unnecessarily.
The six-month rule told me when to look, and thanks to the wait I was able to stand on the shoulders of the early adopters and give it a proper evaluation.
Protect the work.
Put the work at the centre of all your thinking, and opt out of the hype cycle. AI discourse is a tangle of marketing, anxiety, and status signalling, and it's nearly impossible to separate the signal from the noise.
Places like LinkedIn are full of reaction-bait posts, and it's easy to get sucked in and feel like everyone else is on top of the latest developments while you're being left behind.
"Everyone's using it" is not a design argument. Speed is not automatically improvement.
Refocus on your work. If you need to opt out of the spaces where the discourse is at its noisiest, do. The important information — the outcomes that will genuinely help you — will bubble up naturally through colleagues, projects, and the spaces you trust.
Experiment widely. Deploy narrowly.
You should be playing with AI. Experimenting builds vocabulary, shapes instinct, and helps form the opinions you'll need when clients ask the big questions about where AI fits.
But experimentation is low risk. Production is responsibility. Don't confuse the two.
Mess around with image generation. Build a throwaway prototype with an AI coding tool. Use Claude to stress-test a content strategy. All of that is good. That's how you show up prepared.
But deploying AI into work that ships, into client relationships, into products people depend on — that's a different conversation, and it deserves the same scrutiny you'd give any tool entering a critical workflow.
AI as a working assistant
If you’re not using AI at all right now, using it as a ‘personal assistant’ is the lowest stakes, least controversial entry point. Start here and build your understanding, get a feel for what it’s good at, and where it fails.
But remember, it’s an assistant, not a creator. It's the stuff around the work, not the work itself.
Check this email makes sense before I send it. Go back to this project agreement and find where we haven't met a commitment yet. Give me three typical ways this task is approached so I can compare to my own instinct.
I don't come to AI with a blank page. I come with a mess of notes and bullet points and half-formed thoughts, and I ask for help organising them or spotting themes I've missed. The raw material is mine. AI helps me tidy the desk.
This is the part of AI adoption that should feel easy, because it's low stakes. You're not handing over judgement. You're freeing up energy so you can spend more of it on the decisions that matter.
Protect the middle layer – that's where your skill lives
Every piece of design work has three layers. There are inputs: research, evidence, data, lived experience, the brief, the business context. There are outputs: the artefacts, the UI, the copy, the code, the thing that ships. And in between there's a middle layer — the interpretation, the synthesis, the judgement.
The middle layer, that’s design. That's where you decide what the inputs mean and what the outputs should be. That’s where the work is done.
AI is fine at the edges. It can summarise research. It can generate UI patterns. It can write first-draft copy. But the middle layer is where your skill lives, and I think it needs protecting.
When I've worked with AI on something that touches that middle layer, I've felt the regression to the mean. Not bad output — average output. Conventional directions that would work on all projects all the time, but don't stand out for a specific special case.
The kind of thinking that's technically correct but doesn't account for the politics of the project, the dynamics of the client relationship, the thing you noticed in user research that contradicted the brief. AI’s tendency to regress to the mean will limit your innovation.
There's emerging evidence to support this instinct. Anthropic published a study in early 2026, "How AI assistance impacts the formation of coding skills", which found that developers who used AI assistance scored 17% lower on comprehension tests than those who worked by hand. The people who delegated the thinking to AI got the job done — but understood less about what they'd built. The people who stayed cognitively engaged, who used AI to ask questions rather than generate answers, kept their skills intact.
The principle translates and highlights a clear risk. If AI does the sense-making, you lose the reps. Over time, you stop knowing your own craft.
I haven't felt that erosion yet, but it's at front of mind every time I use an AI tool.
Bring AI in when it benefits them, not you
The temptation to use AI is there. There's a peer pressure to it. A quiet voice that says "I bet everyone else is using AI to do this”.
That voice is not a design argument either.
When you're deciding whether AI belongs in work that ships, the question changes from "is this interesting to me?” to "does this make things better for the people I'm serving?”
Some contexts are right for AI. Others aren't. Your client may not be in a place to have the AI conversation at all — corporate governance is slow, and getting permission to use AI tools may add time to a project rather than save it. The practical reality of introducing AI into a client relationship is often more complicated than the discourse suggests.
Sometimes the right call is silence. Not mentioning AI. Seeing an opportunity where AI could help, and choosing not to take it because of the context, the relationship, or the political reality of the project. Reading the room is a skill too.
Your value is in your judgement. Bringing your taste, your specialism, your experience to the work is what you're paid for.
You don't have to chase everything
You don't have to chase everything. You don't have to reject everything either.
Experiment. Protect quality. Keep hold of your judgement. Use AI where it genuinely makes the work better, and leave it alone where it doesn't. The tools will keep changing — they always have. Your job is the same as it's always been: do good work, serve the people you're working for, and know enough about your craft to notice when something's off.
Thoughtful pacing is not falling behind. It's how you've always worked. Trust it.
Am I about to be automated out of a job? Perhaps. I wouldn't recommend a career in digital to a young person right now, and I think pretending everything's fine helps nobody.
But I've been doing this a long time, and so far what I've found isn't threat, it's tools. Useful ones, when used mindfully. The rest, I'm still figuring out. I suspect you are too.
Martin is a designer working on complex digital services and products across government, healthcare, charity, and the private sector.