There’s a moment almost everyone who uses AI seriously has had: the tool produces something genuinely impressive, fast, and free and then, a few minutes later, you spot something that’s just… wrong. A hand with six fingers. A logo that’s subtly not your logo anymore. A sentence that sounds confident but is completely factually inaccurate.
AI is not going away, and it’s only going to get more capable and progressively embedded in how we work. However right now, there’s a gap between what AI produces and what’s usable. The gap in question is being filled by human expertise, life experience and an eye for detail that no AI model has yet replicated.
Even AI agrees with this
Ask most AI tools directly, and they’ll tell you the same thing: don’t rely on them completely. They generate from whatever’s readily available online: patterns, precedents, historical materials but not from lived experience. For example, in our industry, they don’t know what a client wants their brand to feel like. They don’t have taste, judgement or the accumulated instinct honed over decades.
That’s not a criticism of the technology. It’s just a fact about what it is. It means the checking, proofreading and application of real-world sensibility isn’t an optional extra polish; it’s the vital core component that makes the output usable.
The designer who lost work to AI to see it all come back again
I recently spoke to a designer who’d experienced exactly the rollercoaster a lot of creative professionals are living through right now. At first, he lost business as clients decided they could generate their own artwork instead of commissioning him.
After a period of time, the same clients started coming back. Not because they’d changed their minds about AI, but because of what happened after they used it.
“My clients would generate something in AI, love the result, and then discover the output was only screen resolution so it wouldn’t reproduce for print at all. So, they’d come to me to “fix” it.
Some of my other clients asked AI to generate ads and flyers. They liked the overall look, but the fonts and colours didn’t match their corporate brand guidelines and, in some cases, the logo itself had been subtly altered by the AI. Close enough to look right at a glance but wrong enough to fail any real brand check. They came back to me to rebuild it to production quality against the actual guidelines. The client is happy right up until someone points out the problems. Then I’m brought back in, sometimes to fix the specific issues, but often it’s genuinely quicker to start from scratch than to untangle what’s gone wrong in the AI-generated version.”
There’s an interesting psychological distortion here because people had instructed the AI to create something, typed the prompt, made the choices and they took a sense of ownership and pride in the result. That ownership however seemed to make them a little blinkered and oblivious as they didn’t notice issues that, to a trained eye, were significant.
Branding gone wrong
I’ve seen this pattern myself with people creating their own artwork for flyers in shops, social media creative and other branded materials. Suddenly everyone thinks they’re a creative. But you can’t simply replace a creative’s training, natural instinct and years of rich, diverse experience with AI and still expect a campaign to truly land with its audience.
Then there’s the technical details my designer friend spoke about. Peer closer and the creative oddities appear. On top of the visual strangeness, there’s the practical dealbreaker – the AI output is only screen resolution. It can’t be used for anything beyond the web meaning no print, no large-format marketing, nothing that requires real quality.
The self-checkout analogy
There’s a particular feeling that comes with AI remedy work. It’s a bit like being one of the employees in a supermarket who rush to your aid when the self-checkout plays up. They’re there to help the machine do its job, while also knowing the machine is there to replace them in the first place!
It’s a strange position. AI is simultaneously taking work away and creating work – just work that’s less about original creation and more about correction, quality control, and cleanup.
Where this lands
None of this is an argument against using AI. It’s genuinely useful and productive.
That’s probably the real shape of where things are heading: AI is a fast first pass or a tool for narrow, well-defined tasks, with a human still setting the boundaries, checking the output, and applying the judgement that comes from actually knowing the client, the brand, the design requirements and what “looks right” really means in context.
AI and humans working together does work brilliantly. However, “together” is the operative word. Left entirely alone, AI doesn’t know what it doesn’t know and right now, that’s still very much our job to protect.
