Subscribe to industry newsletters

Marketing & Media jobs

  • Account Manager Johannesburg
  • MoreSubmit a jobOpen account
    Search jobs

    Why AI should means 'man loves machine', not 'man vs machine'

    Introduced by Karin Du Chenne, CEO for Insights at Kantar, their latest free talk on why man and machine should be best friends, not enemies, emphasised the fact that we shouldn't fear machines 'taking our jobs' - while none of our jobs will be unaffected by AI, it'll be a case of working with each other, not against each other. Here's why...
    Kantar's AI 'man loves machine' talk.
    Kantar's AI 'man loves machine' talk.

    Du Chenne kicked off the morning by saying if businesses are not making use of mobile as a primary platform to communicate to customers, especially in Africa, they are so last season. But are we adapting our thinking fast enough to what’s available to us?

    How machines are learning

    The first speaker of the morning was Kantar TNS global director of brand Adhil Patel, who reiterated that much of the world is terrified of where AI is going as we know we will be affected by it in some way, shape or form.

    But with the rate of computer power’s growth, AI is only estimated to replace some current positions in 2045, so there’s a long time yet to learn to work with machines rather than against them.

    Patel does not envision of Terminator 2 scenario but rather a symbiotic relationship, where machine aid humans in creating something better than the past.

    Computer power’s constant doubling is what’s driving this innovation.

    Linked to this, Patel mentioned the moral dilemma of self-driving cars, as the robots of the future will have to make some tough choices and dabble in the grey area, despite not having grey matter.

    Autonomous driving is still far away and only in conceptual or prototyping stage at the moment, but given the rate of technological innovation, this will happen sooner rather than later. Have a look at the MIT Moral Machine website for a few examples to consider.

    The digital dilemma – what is digital, what is AI?

    Patel added that the word ‘digital’ is used in so many ways today that it has lost some of the initial power of its definition.

    While various aspects of AI are linked to machine intelligence, most interesting to Patel, in addition to machine learning overall – is the aspect of computer vision and what goes into allowing computers to see.

    We know that vision begins in the eyes, so it’s just optics to start with, but that’s the easy part. The more complex aspects take place in the brain, so beyond just object recognition, computers need to reflect that and add context.

    In addition, Patel mentioned that ‘neural networks’ come to mind for most people when deep learning is mentioned, but just think of the AlphaGo example.

    DeepMind’s AlphaGo is a version of Go – one of the oldest, most complex board games, and AlphaGo beat one of the human Go masters. Then, a new version called AlphaGo Zero beat AlphaGo – and did so by learning the game and its intricacies from scratch.

    Patel ended with a reminder that when incorporating AI into market research, the idea is that AI should do no harm, so ensure that you take good care of all customer data and design for human moments first, as those will evolve over time.

    Modernising surveys and moving to real-time analysis

    The second speaker of the morning was David Wright, Kantar’s Conversational AI Lead, on how the world is becoming more connected.

    More brands are using chatbots to talk customers through the process, while also doing research through product feedback, which is useful in the B2B world as you can engage with the customer on multiple digital platforms.

    Wright said the personal nature of the medium means their bots sometimes even get marriage proposals!

    On the AlphaGo example mentioned by Patel, Wright said it proves the scary speed at which we can shrink the learning process for machines, then use it for brands to better understand the consumer conversation, in-depth.

    We can thereby move from post-project analysis to real-time analysis, with category and tribe chatbots expected to rise as the language model evolves over time.

    Du Chenne commented that this is a huge challenge for market researchers, as much of what they do right now is not chatbot-compliant.

    Helping to revolutionise the way we think about this, Jon Puleston, VP of innovation of Lightspeed explained what needs to be done to bring market research more in line with the opportunities presented by AI.

    Based on Du Chenne’s earlier observation, he said that we need to place mobile optimisation first, and also cut down on survey length as the attention span has shortened, with loading time on mobile of particular importance, as it serves as an exit point for many.

    Puleston said the challenge is to create a survey that people will complete the same way across multiple devices, or else you’re wasting your time.

    For example, ‘dragging and dropping’ to rank doesn’t work well on mobile, so rather implement click ranking. Other ways to optimise screen space include collapsing grids, and the shift to mobile also means rethinking the questions, especially if rating scales run off the mobile ‘page’.

    Another consideration in 2019 is that consumers are distracted by different things at different times of day, and they are likely to complete surveys where they can be easily distracted, like in front of the TV or while on their commute.

    Puleston recommends a more agile approach in designing a survey and being stern in assessing whether the responses to a particular question will add useful data or just serves as a pat on the back.

    He says to also take note of the point where people drop out of the survey or take a long time to respond to a particular question as you may need to change that to minimise the annoyance factor.

    Ultimately, annoyance boils down to repetition and making the respondents do lots of reading and scroll through lists that loop back to the same thing. Make sure your first few questions are engaging and fun, and rather use intelligent routing and filtering to knock out unnecessary options. Also, work on the narrative so the storyline of the survey flows naturally.

    Interestingly, Puleston pointed out that ‘please’ is a trigger word that makes us recognise we’re being asked to do some work, whereas a word like ‘imagine’ is a powerful voluntary construct that may have a better result. He also mentioned the ‘silent dog’ approach, in looking for what’s not being said.

    Puleston concluded that we need to rethink our approach in asking questions, in order to get more truthful answers.

    Man and machine, a magical retail marriage

    Charlene van Zyl, head of customer at Woolworths, was the final speaker of the morning.

    She also spoke of the importance of working with rather than against machines, and of giving them the insights to analyse within parameters in the outside-in approach, rather than inside-out which involves much human analysis.

    Most company Exco meetings are filled with so much data, you’re just adding to the avalanche if you don’t distil it and add in solid use cases, which is where human and machines need to work together.
    The computer is incredibly fast, accurate and stupid. Man is unbelievably slow, inaccurate and brilliant. The marriage of the two is a challenge and force beyond calculation.
    That’s a quote from Albert Einstein himself in the 1940s.

    Offering a slight caution on how we use insights, Van Zyl says that opening up data to scale reveals customer views many brands haven’t before been able to access. But the brand needs time to digest all the elements, and at times use surveys to plug the holes or find out ‘the’ why behind the results.

    Van Zyl also says to check it’s not a highly emotive situation, as customers prefer to talk to a human in these cases. Your brand should implement AI where it makes the most sense to do so. Beware of treating it like a ‘plug and play’ magical box that comes up with answers, too. Instead, it should be seen as a form of partnering as you would with a consultant, for a constant feedback loop.

    Van Zyl mentioned quite a few uses of AI in enhancing human capability but mentioned that the personalisation aspect still offers the chance to annoy.

    Promoting baby products to someone who bought a once-off gift for a friend or chocolate to someone who did a browse-through all the sweet options but ended up purchasing health bars, can annoy.

    When it comes to beauty and fashion, Van Zyl again said to go with caution as a machine may see what size you’ve previously purchased and not realise that it was before a weight loss journey, and could cause offence if offering the wrong size.

    When it comes to customer experience and churn, note that customers may have died or moved out of the household, or their economic situation may have drastically changed. Poor customer experience also keeps them away. Sometimes we just want speed to get in and out without queues, sometimes we want the personal touch. Therein lies the intricacy of being human.

    AI advancements will eventually show more of the ‘why’ but at the moment these show more of the ‘what’ – that’s why humans and machines work best together.

    Du Chenne ended with the words that we need to keep the customer at the centre of all we do.

    As per professor Andrew Ng:
    In the same way electricity and the internet changed everything, over the next few decades, AI will change everything.
    Best we get ready…

    Click through to the Kantar Millward Brown press office and Kantar TNS press office for the latest updates.

    About Leigh Andrews

    Leigh Andrews AKA the #MilkshakeQueen, is former Editor-in-Chief: Marketing & Media at, with a passion for issues of diversity, inclusion and equality, and of course, gourmet food and drinks! She can be reached on Twitter at @Leigh_Andrews.

    Let's do Biz