Nobody sends you a formal letter announcing that the ground has shifted under your profession. It arrives in small comments. A customer wondering aloud whether "something like ChatGPT could knock that out". A stakeholder asking why a deliverable takes a week when "the AI demo did it in thirty seconds". At the company I work for, the message comes from the top and it's framed positively: AI will let us think bigger and move faster. Which is true. But sit with that sentence for a moment as a designer, and you can hear the quiet second half nobody says out loud: ...with fewer of you.

I've heard versions of the "can AI do it cheaper?" question more than once now, sometimes directly, more often implied. And I want to be honest about my first reaction, because I suspect it's the same as yours: it stings. It feels like being told that the thing you've spent your career getting good at has been reduced to a commodity, available on tap, priced per token.

Then I calmed down, and I realised something that changed how I handle the question entirely.

The question is fair, and you need to be able to answer it

Here's the uncomfortable truth: when someone asks whether AI can do it cheaper, they are not attacking your craft. They're asking a completely reasonable question about value. Businesses exist to allocate resources sensibly. If a tool appears that seems to do what you do at a fraction of the cost, of course they're going to ask. You would too. You probably have, about your own suppliers, your own subscriptions, your own tools.

The sting isn't really about the question. The sting is about not having a good answer ready.

Because here's the thing: if a piece of your work genuinely can be done just as well by a text prompt, then the honest answer is yes, it can be done cheaper, and pretending otherwise makes you look like a taxi firm arguing that satnav is a fad. But if the answer is "no, and here's specifically why", you need to be able to articulate that without getting defensive. Most designers can't, because we've never had to. The value of design was assumed. It isn't anymore.

What they're actually asking

When a customer or stakeholder gestures at AI, decode the question. In my experience it's rarely "can a machine replace your taste?" It's usually one of these:

"Can we go faster?" Speed is often the real driver. They've seen a demo where something appeared in seconds and they're comparing it to your two-week turnaround. They don't necessarily want the AI's output. They want less waiting.

"Can we spend less on the parts that don't matter to us?" Clients have never valued every part of the process equally. The exploration rounds, the versioning, the tidy-up: to them, that's overhead. AI looks like a way to compress the overhead.

"Do we still need this role at all?" This is the scary one, and it's worth taking seriously rather than dismissing. Where I work, there are very few dedicated designers outside of marketing. Most business units have none at all. That's not a hypothetical future; it's the present in a lot of organisations. Design as a function is being absorbed into other roles, and AI accelerates that. Which is, frankly, part of why I moved into product management in 2022. I could see which way the furniture was being rearranged. (I've written more about that decision in this isn't my first extinction event.)

Once you decode the real question, you can answer the real question. And your answer gets a lot stronger.

How I answer it now

My honest answer, refined over a couple of years of these conversations, goes something like this:

Yes, AI can produce output cheaper. I say this immediately, because it's true and because conceding it buys enormous credibility. Generating an image, a layout, a first draft of copy: cheaper, faster, done. Fighting this point is how you lose the room.

But output was never the expensive part. What you're actually paying for is a chain of decisions. Which of the fifty possible directions is right for this audience? What should we deliberately not do? Does this design solve the problem we agreed on, or a different problem that's easier to make pretty? AI generates options; it doesn't carry accountability for choosing. When the campaign flops or the product confuses users, "the AI made it" is not a defence anyone accepts.

And I use AI myself, openly. This is the move that changes the whole conversation. I use Claude daily. It does my desk research, drafts my documentation, helps me analyse data, and I've even used it to build working prototypes. So when someone asks whether AI could make things cheaper and faster, my answer is: it already is. I've built it into how I work, and that's why the exploration phase now takes days instead of weeks. Suddenly I'm not the person being threatened by the tool. I'm the person who knows how to wield it.

That reframing, from AI versus me to AI plus me, is the single most useful positioning shift I've made. It's the difference between being the cost that gets questioned and the multiplier that gets kept.

Sell decisions, not deliverables

There's a broader lesson underneath all this, and it applies whether you're freelance, agency-side, or in-house: when a tool commoditises your output, sell your decisions instead.

Every technology shift in creative work has followed this pattern. Desktop publishing made typesetting cheap, and the value moved to typography judgement. Stock photography made images cheap, and the value moved to art direction. Squarespace and Canva made "having a design" cheap, and the value moved to knowing what the design should do. AI makes production cheap at a scale we've never seen. The value moves, again, up the chain: to framing the problem, directing the tools, evaluating the output, and owning the outcome.

Practically, that means changing what you show and how you talk:

  • Talk about problems before pixels. If your proposal opens with deliverables, you've priced yourself as a production resource. Open with the problem, the risk of getting it wrong, and the decisions that need making.
  • Show your reasoning, not just your work. The Dribbble-style reveal of a finished artefact is exactly the thing AI appears to replicate. Walking someone through why, through the options rejected and the trade-offs weighed, is the thing it can't.
  • Make your AI use visible. Hiding it makes you look slow and dishonest when it's discovered. Foregrounding it makes you look current, and it repositions the conversation around judgement.
  • Champion the process, not just the artefact. I'm still a huge advocate for design thinking and UX methodology even though my job title says product manager, precisely because the methodology is the durable part. Research, synthesis, testing with real humans: no model does this for you, and this is where the expensive mistakes get prevented.

What I'd tell my past self

If I could send one note back to the version of me who first heard the "cheaper with AI" question, it would say: stop hearing it as an insult and start hearing it as a brief. The client is telling you exactly what they value (speed, efficiency, results) and exactly what they don't. That's not a threat. That's the most honest requirements document you'll ever receive.

The designers who struggle with this era won't be the least talented ones. They'll be the ones who keep answering a values question with a craft answer, who respond to "why should we pay for this?" with "because look how good it is." Good isn't the argument anymore. Right is the argument. Chosen well, for this audience, for this goal, with someone accountable standing behind it.

AI can't be that someone. You can. That's the answer to the question. And honestly, once you can say it without flinching, the question stops stinging at all.

Want the practical side of this, meaning what to actually build into your workflow? Start with the only 4 AI skills designers actually need.