Why AI Is Too Clever To Replace Machines
Staff Writeron 15 September, 2017 at 01:09
If you’re a marketer, and worried that Artificial Intelligence is about to take your job, eat your lunch and steal your sense of purpose, one session at the All That Matters conference in Singapore brought a touch of comfort.
On the panel was Grace Ho, the CMO and ASEAN strategy leader for IBM, who spoke broadly about the advantages of bringing super-computer data-crunching power to your marketing strategy. IBM, of course, is home to Watson, a high-profile and super-capable cognitive system, capable of doing just that by making sense of mammoth amounts of data and communicating in natural language.
Using Watson, said Ho, “gives marketers the ability to scrape through the internet, through billions of structured and unstructured data. As a marketer, if you can harness all the different sources of data, and make sense of it, this is the key to personalising marketing strategies”.
Grace Ho gave numerous examples to highlight her arguments. In one, IBM was able to help a musician review five years of billboard songs, to understand the ebb and flow of the emotions that under-pinned them. The result was a very successful song that played on the emotion of ‘heartbreak’ – a winning strategy that was gleaned from the data. Not only did this shorten the creative process, suggested Ho, it increased “the chance and probability of a high performance output to the audience you want to target to”.
One other example was of a baker in Hong Kong, who dug through his data to understand why baguettes weren’t selling well on particular days. The answer was that it was raining, and the packaging meant that people knew they’d get wet, so avoided them. The baker was able to build up a responsive baking menu that was integrated with the weather forecast, driving sales and increasing revenue.
Were these good examples of how data-crunching and Artificial Intelligence could improve marketing strategies? Well, yes and no. They certainly helped both the musician and the baker. But could the underlying insights have been realised without something as awesome as Watson?
Another member of the panel, Fionn Hyndman of Outbrain, spoke glowingly of human frailties. “Humans are fundamentally flawed in judgement,” he said. “AI will give us better answers in many cases, because of the psychological biases that we put on decisions”. But, he suggested, our human flaws gave us a paradoxical advantage in reaching some conclusions, simply because we had the flaws that led to intuition and judgement rather than pure thoughts.
In short, while AI might help us refine the insights and translate them more effectively into strategy and action, don’t remove people from the process. They might just come up with the insights that long pieces of bread don’t fit well into bags, and so get wet in rain, or that songs about heartbreak will resonate with teenagers.
As Grace Ho puts it herself, “humans and machines will co-exist… Machine learning will not develop a human’s sixth sense, but will assist us in making better data-driven decisions.” Marketers, your jobs are safe, but remember to be humans and let computers do the tricky stuff.