Marketers, beware these six misconceptions of AI
The subject of AI is mired in hype.
AI proponents point to utopian futures where computers serve humanity and help us to achieve immortality while sceptics worry about the existential threat of computers choosing to make humans extinct!
The reality of AI as it stands today is quite different from what has been predicted in popular sci-fi. From a marketing point of view, it is fair to say that AI will permeate all aspects of the profession to some extent. That means that in general, our marketing should get smarter.
Econsultancy Founder Ashley Friedlein wrote earlier this year that there is already a lot of AI, or at least machine learning, in the technology we use for marketing. AI applications like machine learning will mean that chatbots and virtual digital assistants will play a much more significant role in customer interactions for the foreseeable future. Other applications include intelligent email subject line optimisation, AI-driven ad tech, content that is optimised using AI, social media optimisation, smart pricing, intelligent product recommendations, image/speech recognition and so on.
That’s all great but there are also a lot of misconceptions about what AI can and can’t do that are worth examining. In anticipation of Econsultancy’s forthcoming update to our 2016 report – Marketing in the Age of Artificial Intelligence, I thought I would consider some of the popular misconceptions about AI.
Misconception 1: The machines will rise and take over the world
AI is not going to take over the world any time soon. Jerome Pesenti, a leading pioneer in AI who was involved in the development of IBM Watson AI, speaking at Web Summit 2017: “If you hear people with this doomsday scenario…they don’t know what they are talking about. Anybody who has any practice in the field of AI knows that we are not yet there”.
Misconception 2: AI is all hype
Okay, so AI won’t become sentient anytime soon but that does not mean that it is all hype. At Web Summit, Jerome Pesenti referred to two important innovations in AI to demonstrate.
CAPTCHA: Everyone is familiar with CAPTCHA, an acronym for “Completely Automated Public Turing test to tell Computers and Humans Apart”. Basically CAPTCHA is used on websites to distinguish human from machine behaviour, typically as a way of thwarting spam.
Letters and numbers are rendered and jumbled together in different ways so that only humans can read them but they are too difficult for a computer to recognise. I find it difficult to identify the letters in a CAPTCHA at the best of times.
The ability to crack CAPTCHAs has become a key benchmark for artificial intelligence researchers. In a research paper published in the journal Science in October, scientists report a new model that can break a CAPTCHA. According to the paper, the model “was able to solve reCAPTCHAs at an accuracy rate of 66.6% …, BotDetect at 64.4%, Yahoo at 57.4% and PayPal at 57.1%.”
You might be thinking so what? Well, the point of this research has nothing to do with CAPTCHAs. It’s about making robots that can visually reason like humans. The long-term goal of this research is to build intelligence that works like the human brain. CAPTCHAs represent a natural test for examining whether a system can work like the brain. This suggests that we may have to take it for granted that computers will be able to work in some ways like the human brain.
Facial recognition: Facebook has created a system that with enough pictures, can learn to recognise your friends better then you can. The system works so well that Facebook was pressured to close the feature in 2012 after concerns were raised from regulators and privacy advocates. In April 2018, Facebook said that it would bring the feature back by asking users for permission to identify them in photos and videos.
So what? The system helps Facebook offer friend suggestions. When new connections are made, users have more reason to spend longer on Facebook where they can see more ads etc.
Misconception 3: AI will take all our jobs
It is true that AI has the potential to change how we work. But maybe it is an over simplification to say that machines will take our jobs away. In the short-term, companies are generally evaluating whether AI can augment human activities, enabling them to work in newer and smarter ways rather than replace them.
For example, Dutch airline KLM is using AI to enhance the customer journey whilst maintaining a human touch. KLM is using bots to push boarding passes, status updates and some other messages. They are also using AI to support agents with suggested answers but they are nowhere near standalone conversation machines.
KLM’s Karlijn Vogel-Meijer at the Festival of Marketing
Chatbots have a future but it is extremely complicated to create human like engagements with real people for any great length of time. Chatbots will never be better than your best customer rep. They can’t replicate deeply human things like empathy. According to KLM social media manager Karlijn Vogel-Meijer, speaking at Festival of Marketing, KLM won’t automate its customer support function. Karlijn categorically stated that KLM wants to “to deliver a personal feel, and we truly believe bots are not able to do so at the moment.”
AI may well gain prominence in tasks that can be codified such as certain administrative processes, reporting and data analysis. However there will always be a need to manage and maintain customer relationships. Work that involves a human touch will not be automated. From a marketing point of view, this includes design, storytelling and that most human quality, empathy. This is certainly work worthy of a marketer’s brain.
Misconception 4: Digital advertising can be automated
In advertising, AI solutions can be used for automated scale, speed and insights in areas such as campaign optimisation and audience discovery. But according to Alex McIlvenny, UK Country Manager at native advertising platform Ligatus, AI doesn’t work for programmatic in its current form.
“AI works well for systems like Amazon’s Alexa where a single ecosystem is controlled by one platform and the computer can learn to make its own decisions without human input, but it isn’t effective for the multi-platform environment of programmatic advertising.”
The programmatic ecosystem is characterised by multiple platforms that are plugged in to each other. According to McIllvenny, before AI can be used effectively in programmatic “we need to find a way for platforms on multiple ecosystems to talk to one another”. He notes that even once that point is reached, human input will still be required rather than relying on AI to completely automate campaigns.
Alex McIlvenny, Ligatus
“AI should not be fully replied upon to make the best advertising decisions and will never be able to go it alone in programmatic. Consider the subject of brand safety, which is a highly nuanced concept AI may never completely understand. Machine learning algorithms can find the ideal user to deliver an ad but they may not be able to evaluate whether they are reaching that user in an appropriate and brand safe environment. We’ve already seen brand safety issues arising on the larger media platforms where algorithms run with little human input, resulting in ads being served alongside questionable or damaging content.”
Misconception 5: AI adoption is all about technology
That’s not correct. If AI has the potential to disrupt jobs, marketing leaders will need to be mindful about how to introduce AI to their organisation. AI adoption is about culture which requires careful consideration.
At a minimum, any modern marketing organisation needs to be responsive and agile. This requires providing marketers with access to learning and development to ensure that they are open to change and further, to empower them to think about how AI can augment what they do rather than replace what they do.
Misconception 6: It’s called AI so it must be AI
AI is a field that has long been crippled by a cycle of inflated expectations and disappointing results. In fact, many of the examples that are commonly provided to demonstrate AI are in fact brute-force data analytics. Consider chatbots, they can’t do open interactions. Rather, they can do scripted demos before passing the user off to a real service agent. We’re all marketers though. Let’s be honest, AI sounds better!
Alex McIlvenny of Ligatus suggests that as AI continues to develop, the marketing industry will move towards what he called Human Aided Artificial Intelligence (HAAI). “I believe this will be true of most industries where AI has an impact on work practices or employment”.
Whatever way AI evolves, marketers would do well to consider Amara’s Law: “We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run“.
That means that it will be difficult to predict what how AI will develop and the impact it will have in the long run. In Econsultancy’s How Marketers’s Learn report we wrote that the only inoculation against disruption is learning. Learn about AI. Learn about its potential impact on your industry, your organisation and your job. And then focus on broadening your skills so that if AI should impact your job, you will be ready to respond.