If an A.I. machine can optimise an ad to an almost infinite degree, does this mean all ads will end up looking the same? There can only be one ‘best’, right? We explore this A.I. chin-stroker…
On 22 May, we welcomed Tom Ollerton, acclaimed UK tech innovator and podcaster, to Indicia Towers in Bristol. Tom gave a fantastic, thought-provoking talk about the creative potential of A.I.
Afterwards, the floor was opened for comments. And one question that captivated the room and caused quite a debate, was this
If all brands use creative A.I. – technology that fine-tunes creative output to cognitive biases and selects every tiny detail to drive response – will ads become homogenous in their pursuit of effectiveness?
Indeed, this seems a plausible suggestion, and a worrying one for us marketers, who are focused on the range of creativity that we see and create, and on differentiation for our clients.
Though we have no clear answers about what the future of A.I. ads will hold, let’s explore this question.
Clone ad nightmares
Within product categories, many products are essentially identical. Take the label off a Colgate and an Aquafresh tube of toothpaste and what’s the difference? So, if you ask A.I. technology for creative about similar products with similar price points and for similar audiences, who’s to say the outcomes won’t be incredibly similar, too?
Here’s a quick example: Coca-Cola and Pepsi are both creating a social ad. As similar products, you could imagine that each brand’s A.I. technologies would pick comparable photography – say a can placed in a hand, with beads of fresh-out-of-the-fridge condensation on the side, performed best in summertime. Say a centre-left product placement and a top-right headline performed best of all placements. Perhaps ‘refresh yourself’ was the highest performing CTA.
What would the result be? Two robot-designed ads that are barely distinguishable?
Theoretically, perhaps. But this scenario doesn’t take into account a few factors that leave room for differentiation.
Data sets
The output of A.I. technology differs depending on the brand using it. This is because A.I. works from the data sets that each brand can access. And of course, every brand has different customer data.
Though Coca-Cola and Pepsi are similar products, they hold different customer and prospect data. People tend to consider themselves Pepsi people or Coke people, and these groups will differ somewhat in regards to attitudes and demographics. What engages one type of customer will be ignored by another.
As a result of the differences in data, the ads should therefore differ. Especially when you take into consideration that A.I. can create a near limitless number of ad variants, tailored to narrow customer segments.
The implication is clear – the battleground becomes data. Brands with access to the smartest datasets will win out.
Brand guidelines
As well as the data sets that each brand has access to, brand guidelines will have a big influence.
Having different parameters set by the brand in regard to colours, shapes, positioning and language, means ads can’t become all that alike as they’ll have unique identities to work within.
Eventually, even brand guidelines could be machine learned. But in the near term, the grey area is created by human involvement – brand storytelling is still a vital part of the creative process. Inspiration is still a human skill. But we can’t assume machines won’t do it in the future.
Growth and adaptability
Ultimately, creative A.I. is still in relative infancy. We have little idea how sophisticated it will become. But it’s predicted that, in the not-so-distant future, it will be capable of independent creative thought.
Just as the advertising greats invent creative ideas that captivate – it’s entirely probable that A.I. will be able to think outside of the box, too.
In this case, ads will differ from one another as A.I. will be able to spot people becoming immune to certain kinds of creative. In reaction, they’ll adapt and create something new.
Scary? Exciting? We say, it’s a bit of both!
.png?width=82&height=41&name=Vector%20(4).png "Vector (4)"){kind=small}
