Ever had that sinking feeling when someone showed you a piece of work to feed back on, or sign off? You want to love it, it might be on-brief, with no glaring errors, but it’s just - meh. You look through it again, searching for clues, hoping something will stand out. Probably, the best you can muster is ‘I’m not sure it’s quite there yet.’ The least useful bit of feedback ever, but we’ve all said (and likely heard) it.

Philosopher Michael Polanyi coined the phrase ‘tacit knowledge’, for the idea that we know more than we can explain. My ultra-talented design director colleague Simon Anderton is obsessed with this concept. The way you can pick out a friend in a crowd of thousands without being able to explain how. You just ‘know’. It’s the same as when work falls flat with no justification. Your instinct understands something reason can’t explain.

So, what is it? Researchers at the Max Planck Institute studying the effect of music on our brains found that pleasure came from the interplay between expectation and surprise. A startling chord, or a familiar passage resolving in an unexpected way. So pleasure, perhaps, can be found in the gap between what we expect and what we get. I can’t prove it (yet), but I suspect that ‘meh’ is what’s left when the work does exactly what you knew it would, when there’s no gap for your brain to reach for.

And now, to the crescendo. There’s a reason so much work does exactly what we knew it would. It’s being made with the same tools. In 2024, Anil Doshi and Oliver Hauser gave some writers story ideas from an AI and left others to their own devices. The AI-assisted stories were rated better written (especially those by less creative writers). They were also more similar to each other. The authors called it a social dilemma: individual writers were better off, but together they produced a narrower range of stories. A year later, a team at Wharton asked people to invent a toy from a brick and a fan. Among those using ChatGPT, 94% of the ideas overlapped, and nine of them - without conferring - independently named their toy ‘Build-a-Breeze Castle’.

If you work in brand or marketing, this is familiar ground. Better and different are separate measures, and AI tools are very good at better. But when every team in a category draws from the same well, the most predictable work becomes the most referenced. We can all feel it. Across seven experiments, Colleen Kirk and Julian Givi found that when people believed an emotional marketing message had been written by AI rather than a person, they were less likely to recommend the brand, and less loyal to it. The effect faded when the message was factual, or when the AI had edited a human draft. People seem relaxed about machines handling the facts. What unsettles them is an ‘unsecured’ feeling.

So where does difference come from? There’s a hopeful thread in the research, and it’s about timing. Wharton’s Kartik Hosanagar and his co-author tested different ways of writing with AI and found that keeping people in charge of the early ideas, and bringing AI in later, preserved far more variety. The study is still awaiting peer review, but it fits the rest of the evidence – we know AI chooses the most probable idea, and it’s harder to differentiate when the starting points are all the same.

That messy beginning is where difference comes from. The varied personal recollections, the wrangling of thoughts and shapes and memories, half-remembered conversations, different cultural associations, fresh perspective; your tacit knowledge.

None of this is an argument against using the tools. But it is an argument about using them too early. By all means use them to sharpen, test, edit and check. But rely on yourself to add that initial spark of direction. And teach it to those new in the industry who – without time and space to develop their own tacit knowledge – are at risk of never (or, horrifyingly, only) feeling the ‘meh’.

Next time a piece of work leaves you cold, ask yourself if it’s because you were given exactly what you expected. If ‘not quite there yet’ might really mean ‘this is already everywhere’.