Posted on 21 September, 2026

Why use a human for design any more?

a white mannequin wearing a white mask

AI in the design space

This has to be the most pressing issue in the industry right now. Designers are struggling with an existential crisis of automation caused by ‘AI’ but I’m not sure the take over is certain, or that the models we’re competing with are as competent as Anthropic or OpenAi would like us to believe (because hey, they need more cash!). Plus, there are so many other advantages to working with a human…for starters I’m currently not wasting thousands of gallons of drinking water or polluting my nearby rivers.

Time to justify my job

A designer handles hundreds of complex interactions for each project, communicating with a client is the foundation of good design. It’s not prompting, then re-prompting, then prompting again to get what you need. A good designer listens, writes a brief, makes the edits and takes responsibility for the work. Designer and client the work in tandem to get the right results, it feels weird to be extolling the virtues of communication vs prompting…but here we are.

And it’s the little requests that make the relationship click, the ‘can you just?’ or ‘this needs fixing, can you help?’ Requests that are all vital parts of projects, large and small. A conversation is just more nuanced than writing a prompt, ideas bounce, vague suggestions are unpacked and then realised. The whole thing is just way more satisfying.

Relationships with clients are built over time, there’s a shorthand developed that makes them feel comfortable sending all manner of requests. They’re trusting we can understand and then create what’s required. All this without making Elon Musk any wealthier.

Understanding your business, not just your prompt

There’s an online consensus that prompting a machine trumps talking to a human, every time. But there’s two things I’ve started to notice about using ‘vibes’ to create that are undermining that stance. Firstly, the suite of AI tools available has become - and is becoming - ever more complex. Maybe you need one AI to write your prompt, then another to create the images, then another to animate, then another to turn this into a website…etc, etc.

For many end users this is time and money (tokens aren’t ‘cheap’ when you multiply them across multiple platforms!) they don’t have, or they don’t have the patience to endlessly re-prompt in the hope the the machine gets it right this time.

Secondly: once we start to move on past the initial wow that our computers can generate images and text, we’re often left with something that’s not fit-for-purpose, or is probably lifted from unlicensed source material and is going to make a printer call in a fit of rage and start shouting to you about 300dpi, CYMK and vector NOT raster! Yeah, that guy on YouTube is cranking out websites like cheap fashion, but you’re not getting the same results. Mayne this thing is a all hype no trousers…?

Final thoughts

Don’t get me wrong, ‘AI’ has it uses for creatives and clients. I had some poster ideas sent to me from for a client, generated via ChatGPT, as part of a brief and they were great. They really helped steer the project and clarify what they wanted. Guess what though…? There was still plenty to do to get them to press and looking top quality.

I’m not about to (sadly) cancel my Adobe subscription and run for the hills…yet. As long as people was to call and talk about their ideas, I’ll listen and start ‘pushing pixels’ until we’re both happy. It seems to work and might be a calm oasis in a desert of AI slop and LLM hallucinations.

No AI was used in the creation of this article.

+44 (0)7540 815 793