With the rise of martech, automation and AI-led customer engagement platforms, data has never been more important. But where does that leave creativity?
Despite experiments with AI-creative, from Google’s Magenta composing music to deep learning platform Vincent turning sketches into Van Gogh, human intelligence still has the upper hand in this field. No matter how complex the algorithm, there is (for now) no machine equivalent of serendipity, that unexpected moment when inspiration strikes and a truly original idea is born.
The longstanding debate of creative vs data aired again during ADMA’s recent Digital Think Tank and Town Hall, Marketing in an AI-First World. But it’s not a binary outcome – we clearly need both. The data that fuels the machine should, in turn, give us more time to think, spark new ideas and be creative – because we no longer need to spend time on manual, rote tasks that could be automated.
“We’ll get our lives and our brains back,” said Beyond Intent cofounder Aryeh Sternberg. “I think there’s a natural blend between what machines can do – volume data processing, insight and automation – and what humans can do. But I haven’t yet encountered a technology capable of creativity anywhere near approaching a human. “
Creative is the clear point of difference
We’re not in a level playing field yet when it comes to data. But when every major brand is tracking the same demographics, behaviour, previous purchases and preferences, advertising messages could become homogenous without some creative intervention.
“The intelligence of the machine is actually only 50 per cent of the equation – you still need content that's going to be engaging enough to really deliver that cut through,” explained Mike Handes, Customer Success Director Marketo, in his address to the Town Hall.
Algorithms can’t predict chemistry
Despite all the data available there is still no single formula to a top-rating TV show, according to Foxtel’s Group Head of Digital and Marketing, Lifestyle Brigitte Slattery.
She explained how an algorithm might have missed a trick on Foxtel’s show Love it or list it. “If a machine analysed the best format for that show around the world, it would have told us to use a male and female host.” But the chemistry between the two male hosts has made it a hit.
Similarly, Netflix thought it was onto a sure winner with its original series Marco Polo – drawn from the same plot formulation and genre as Game of Thrones. “It was cancelled after season 2,” noted Brigitte. “On paper, Netflix has access to the insights, the algorithms, the best producers, actors and screenwriters. That still doesn’t guarantee a hit series.”
‘Surprise me’ is a human desire
The more data platforms like Facebook, Netflix and Amazon collect on us, the more personalised and relevant their content response or recommendations can be. But is that limiting us as consumers from the unexpected delight of a new discovery?
“Marketers still need to be able to surprise and delight,” said Mike. “I think sometimes you need to incorporate some random elements into the algorithm. Some retailers do this just to not appear too creepy.”
Automation still needs some hands-on input
Machines don’t create the algorithms – humans do. And we still have a responsibility to guide a bot’s moral compass – because as Microsoft’s chatbot Tay showed, when machines learn from online sentiment they can produce racist, misogynist, anti-Semitic hate speech to rival the internet’s worst trolls. Humans will still need to maintain control over those ethical parameters.
So while human imagination and empathy still matters as we move into this AI-first marketing world, it’s worth remembering just how subjective creativity is. As McCann Japan’s new AI Creative Director proved, bots can already be both creative and funny – with a robot-directed TV spot for Clorets mint tab preferred over it’s human-directed equivalent by advertising executives.
For now, however, we’re seeing machine learning achieve great results in testing, targeting and automating – leaving the storytelling and creative strategy to people. But don’t get too comfortable – a bot might still be capable of doing your job in the very near future.