Where do AI ideas come from?

“Average is easy to find with AI,” a speaker told the recent CIM London event on AI and marketing in 2024. There was applause.

What does that mean? Is it true? As marketers, should we want it to be true? The last thing anyone wants is that their work is bland. Being average renders marketing content invisible.

Working with AI, many creative teams are disappointed with the results. To the non-specialist, the creative looks great, but to professionals, it appears utterly mediocre.

Submit a standard ad brief to a sophisticated AI model like GPT-4, and you’ll receive a credible, plausible response. However, such campaigns rarely make judging shortlists and, as pitches, often rank as the less favoured “second place” to more vibrant, human work.

The comfort from this story is that humans are irreplaceable, that our most sacred quality – creativity – is something that still cannot, and may never be, replaced by machines. It’s reassuring and feels true.

It’s wrong.

Typical homework for Stanford Design School students used to be generating many ideas for a design challenge. Now, a year after OpenAI’s launch of ChatGPT, the assignment is more likely to be using AI to generate lots of ideas by asking interesting questions and selecting the best. This is not ceding creativity to AI, it is realising its true strength – a cognitive booster engine that we can use to go further, faster with our human minds.

Spotting the outstanding idea is the uniquely human part of the creativity process. Knowing what will stand out, having the taste and instinctive sense of what is brilliant work over quotidian also-rans.

One of our clients wanted to understand how the best brands manage to consistently produce the best ads, even when they work with similar agencies to the dreaded “average”. Training people how to have great ideas didn’t work.

We built them an AI bot – a much quicker and simpler process than it was even three months ago – that could look at any content and reverse engineer it, writing a plausible version of what the briefing document was. The AI would then explain the process to develop the work, and at which stages the creative idea could be compromised by bureaucracy or threatened by decision-makers playing things too safe.

We processed inspirational work from other brands first. The bot created eerily familiar stories of how the brilliant ads were never made.

Then the client tested their ideas with the same bot and it predicted the battles they would need to pick to make sure the work was outstanding.

Limited experience of working with AI leads to average questions being put in and average responses coming back. Thinking creatively about how to work with AI will yield the best results and allow creative minds to reach even higher. 


Author: Antony Mayfield, CEO of Brilliant Noise