We’re sitting comfortably in the age of Amazon’s Alexa, Google Home, and Apple’s Siri. And what do they all have in common, apart from the fact that you can shout commands at them and they’ll do their best to meet them? Well, that’s part of it.
Alexa, Google Home, and Siri could be classed as Artificial Intelligences. Yes, the pesky AIs the late physicist Stephen Hawking once warned would end the human race. Tesla CEO Elon Musk also jumped on the bandwagon. At SXSW, Musk said: “Mark my words. AI is much more dangerous than nukes, so why do we have no regulatory oversight?”
But despite the doomsday press surrounding AI, The Drum reports that generally, digital marketers are optimistic about including AI in their futures.
If we think about it, AI’s huge potential for marketers in the future lies in its ability to work repetitive tasks. And if AI can handle the ‘grunt work’, then it leaves more room for marketers to focus on content strategy and creativity in a bid to boost SEO and the link building network of their company, and that company’s clients.
This is further backed up by the Forrester study, which concluded:
We think perhaps the most stunning finding that the survey shows is that most marketing firms agree or strongly agree that AI will reinvent the retail industry (88%) and dramatically change what their companies do (81%).
Maybe if we work hard enough to shove the latest episode of Black Mirror into the back of our minds, we can focus on how technology can be warped–for good. For most of us, like I mentioned above, AI is already an active force in our home. Alexa will help you with your shopping list, and Siri will assist you in a voice search. Whether we like it or not, it’s becoming more of a commonplace feature in the lives we live today.
So, surely, utilizing the benefits of AI and exploiting how they can vastly boost efficiency within our business is a good thing? Right?
Let’s just try and break it down. AI wouldn’t have gotten this far if there wasn’t some purpose to it. So why and how is it so appealing to businesses? How can it improve business opportunities? And what are some fears the business world and/or the general public have in regards to AI?
Why is AI so appealing?
Previously, marketers hesitated in incorporating artificial intelligence into any business plans or strategies. But recently, we’ve witnessed a huge shift in both confidence and reliance in the digital marketing world on artificial intelligence.
Well, you’d assume that any business with a decent revenue would not want to risk working with an ambiguous system. But the truth is, the actions carried out by the AI is ultimately programmed by a human being, initially. Yes, there runs a risk of misunderstanding, or perhaps the programmer’s honorability is not much to be desired.
However, if we consider the benefits of a well-oiled machine; a successful machine…Then yes, the digital marketing world’s upheaval in the AI direction is very understandable. The clear, direct and usually accurate results an AI gives is desirable, and most sectors are likely reaping the rewards of those results now.
One of the most obvious reasons for utilizing AI is efficiency. If it took an employee an entire day to comb through a customer’s purchase and search history, as well as actual purchases, and then use that information to recommend suitable products for future purchases, that’s impressive.
But if an AI can do that accurately, and in a fraction of the time it takes manually, then surely you’d opt for the AI. When you think about common marketplace site (Amazon, eBay…) recommendations, they aren’t enforced by an employee combing through your search history. They’ll have collected data from you based on previous site activity and made use of artificial intelligence to predict what you’re likely to look for.
It’s basically why, after you’ve bought a pair of headphones, for example, Amazon will recommend similar items like speakers or headphone cases to push for another purchase.
AI can really boost potential target areas in terms of digital marketing. Right now, we could be seen in the baby-stages of AI, despite advanced strides made by big-hitters such as Amazon, Google and Apple. So, aside from the increase in efficiency due to the automation of websites and their systems, why else might 2018 be the year of the AI?
Greater Personalization of User Experience
Content may be king for the digital marketer, but for the business as a whole, the customer is king.
The notion of living in a surveillance state and all the reports about information collection might well be very true and very scary. But whilst collecting data in such a massive way is politically divisive and certainly a hot topic, it’s also business.
No online business will deny the fact that they most likely take note of your search history on the site, essentially keeping an eye on what you do on their page. It’s probably most obvious on popular shopping sites like Amazon and eBay.
Immediately noticeable on their homepages are features such as recommendations based on your previous purchase, your search history appearing as you type in the search bar, and a section on ‘what other customers bought…’ along with the item you intend to purchase, as you’re about to buy it.
These techniques are usually quite effective. For example, if you’re not too sure about what headphones to buy, Amazon recommending you similar items to the one you’re viewing could be very helpful. Further recommending that you purchase a headphone case as well will push sales on the business side, because it’s a sensible recommendation that you might not initially have thought of. So, the likelihood of you clicking on the headphone case and viewing it as a potential purchase is high, rather than if you received a generic advert or sales push.
In a way, the AI ‘knows’ what you might need in addition to your potential purchase. The process of the AI making an assumption about what might happen (you buying the headphone case too) is called predictive analytics. And one step up from that, the AI finding a way to make it happen (e.g. via Amazon’s suggestions or ‘another customer bought this…’) is called prescriptive analytics.
Usage of chatbots
Other ways AI can personalise user experience includes the use of chatbots. By using chatbots, AI can help the website direct you precisely to the content or link you want. It’s almost like ringing your doctor’s surgery in the morning.
Your receptionist will always ask you your symptoms and what you need the appointment for. They’re like a filter before you see a doctor. If the symptoms you describe are mild, they might direct you to a pharmacy for over-the-counter medication. If the receptionist thinks your case does require a doctor’s attention, they’ll set up an appointment.
That’s essentially how a chatbot works, too. By asking a series of set questions, the AI can create an image of what you’re looking for, before directing you to either an online consultant or a specific webpage. This eliminates the need for you to search around a website.
Because information is so readily available now, a site that requires a lot of clicking around is a lot less attractive than a site that offers you the answer to your query quickly and accurately. Therefore, usage of chatbots will not only improve user experience, but it’ll benefit the business by potentially increasing conversion rate.
Usage of augmented reality
Another employment of AI that’s maybe further out on the horizon is the use of augmented reality (AR). AR is defined as an “An enhanced version of reality where live direct or indirect views of physical real-world environments are augmented with superimposed computer-generated images over a user’s view of the real-world, thus enhancing one’s current perception of reality.”
As AR technology develops and becomes more widespread, it is definitely an attractive use for it. In theory, it’ll ease the decision-making process from the customer’s perspective. Effectively, it’ll increase your business’ revenue by employing AR. The more informed the customer feels about a product, the more confident they’ll feel in assessing whether or not what you’re selling is right for them.
If you wanted to purchase an item, experiencing it via AR means customers can ‘feel’ and see the product before buying it. Apart from being a much more personal experience, it goes a long way to instill trust in a company’s product if the customer can ‘try’ it before they buy it. It’s almost like taking a car out for a test drive to see if you enjoy the feel of the car, rather than just reading up on an extremely detailed specification of the model.
Removing the guesswork from the world of digital marketing
Instead of the cumbersome process of never-ending market research and, let’s face it, a bunch of guesswork in order to tap into your ideal client base, AI can focus on delivering its adverts to specific people.
If you do your weekly supermarket shop and you buy avocados all the time, your supermarket’s website will have that data logged. The next time you shop online, there’ll be a section that prompts you to buy your ‘favourites’–because that site knows you are a regular purchaser of that product.
The same principle can be applied to the world of advertising and digital marketing, too.
The proposal is to have a digital campaign–for example, an online advert–tailored to each customer. It’s exactly what it says on the tin. By tailoring an advert to a customer, that advert is always going to be relevant to the customer’s interests. This is a big step up from the generic, blanket coverage offered before.
Before, marketing could be seen or even accused of being disruptive. Advertisements or sales-oriented online content were both ill-reputed and did not provide a strong return on investment.
Billboards were designed to be eye-catching and distracting, in order to pull the customer’s attention away from what they were already doing. None of these were designed with the customer’s interests or tendencies at the forefront of the plan. Take a billboard for example. A strong marketing campaign would be disruptive and distract enough people to be successful. However, it wouldn’t appeal to everyone. Companies could guess what passers-by would be interested in, but a level of uncertainty remained.
If we applied this to online marketing, the picture is very much the same. Arguably, Google has gotten smarter over the years. In terms of recommended websites or related searches, it’s heavily reliant on AI, which in turn is heavily reliant on data collection of search histories.
By analyzing all this data, the AI might figure out that if a customer searched for ‘mountain bike’, they would go on to also search for ‘mountain bike trails’ or ‘bike helmets’.
The evidence of Google and ‘intelligent learning’ is evident. Type in pretty much any term and Google will not only give you the most relevant and reliable results to answer your query, but it’ll go the extra mile to provide you extra information via related terms others have searched for. If you Google a film title, it won’t just cough up the IMDb page. You’ll get a cast list, a synopsis, etc.
On a larger scale, Piccadilly Circus has already implemented AI-driven advertising. Based on a person’s gender, height, facial-recognition-determined-age, among other factors, the advertisements lighting up the famous board will be specific to the passer by. So, if you’re a six-foot male in a well-cut suit, an advertisement for a luxury watch might flash up as you walk past.
Of course, not everything about this approach is perfect. However, despite how complex it sounds, it is only truly reliant on one thing: information. And that is common if valuable, commodity online. Every page we visit on our laptops, every purchase we make online, every search we make, every social media post we put up–it’s all logged.
The more information is freely available online, the more we’ll use the Internet to grab that information. Therefore, every time we do that, the Internet will use us to curate a more accurate image of each user.
It’s a very detailed but ultimately efficient silent exchange.
We’re still a long way from the day an AI can practically read our minds. You might randomly fancy a banana milkshake halfway through reading this. That doesn’t mean Google will suddenly sense that and automatically order you a milkshake via Just Eat or something. But undeniably, using AI in the world of digital marketing is becoming more commonplace.
Effectively, it’s plucked the ambiguous grey area surrounding some client personas. Even though we know we might know the general audience to target, a digital marketer cannot compile every user’s data and know when to reach for specific points collected.
If the AI were a personal shopper, its predictive analytics would be based on the assumptions it’s made of the historical data it’s collected from you. In other words, your behaviour. The dialogue would be something like, “So you’ve told us that you’re 41, a size 16, and female, on sign-up. Via your browsing history, we know your price range, the colours you like, the type of clothing you’re looking for, and the items you’ve viewed the most. Based on all of that, we think you’re looking for a patterned sweatshirt.”
The next step would be the usage of prescriptive analytics. As we’ve talked briefly about before, this is essentially the action the AI would take after assessing the information collated. Back to our personal shopper, this is the stage where you’d be told: “What I’m going to do is recommend this patterned sweatshirt within this price range you’ve specified. But here’s some other similar items that you might like too, because you’ve looked at or purchased these items before, or because other people who have the same taste as you have also bought.”
What are the fears surrounding AI? Are they warranted?
AI and the collection of online information have been widely debated across multiple sectors. Most famously, Edward Snowden was made an instant celebrity after he leaked intelligence on the United States’ surveillance systems. Afterwards, the fear of ‘being watched’ in an almost Orwellian state became pretty much synonymous with the word ‘intelligence’.
So add ‘artificial’ in front of it, and suddenly we’ve gone from a hacker who can remotely turn your laptop’s webcam to empathising heavily with Will Smith in ‘I, Robot’.
The questions whirling around the topic of AI are never-ending.
Are they truly removed from bias? Just because there isn’t a human analysing the data with a particular political opinion, if AI is so intelligent, can they form their own political agenda? Or, more relevantly, if an AI requires human input in the first place, in terms of programming, is biased truly non-existent when it comes to AI?
What about public perception? Charlie Brooker’s anthology series ‘Black Mirror’, described as a set of “sharp, suspenseful, satirical tales that explore techno-paranoia” has proven hugely popular.
Long story short, it features standalone episodes that cleverly uses a hypothetical–but not too out-of-this-world–technology or artificial intelligence to serve as commentary on humanity.
With the popularity of the series, it’s not difficult to see why people are wary about the amount of information we share on social media and how reliant we are on it; how technology can be wielded by people in power to deceive us; how the ease of using technology as a crutch can conversely be crippling.
Yes, it’s fiction, but when reports seep out about how the show isn’t, in fact, so far from fiction at all, then it’s incredibly easy to connect the dramatised points of the stories to ‘what might happen in real life’.
Tay’s accidentally racist agenda
Perhaps the richest example of how AI can go wrong is when Microsoft’s chatbot on Twitter, Tay, was deactivated hours after launch for making racist comments. It’s maybe the leading case of how, despite intricate programming and automation, human error and an ‘independent’ technology can still cause a recipe for disaster.
Innocently, Tay started out as a chatbot designed to learn more via interactions with people on Twitter. And as Microsoft hoped, a way to engage with millennials who are the ones primarily active on the platform.
But as is the case with social media, things did not go as planned.
After a series of tweets that included the assertion that Ricky Gervais had learned totalitarianism from the notorious dictator Adolf Hitler, as well as tweets about Trump, users were beginning to notice the failings with Tay.
If you want an example of just how badly AI can go wrong, perhaps you need to look no further than this tweet:
bush did 9/11 and Hitler would have done a better job than the monkey we have now. donald trump is the only hope we’ve got.
Interestingly, Microsoft claims that Tay is actually a combination of AI as well as human, editorial staff. It does seem that in most cases, Tay was repeating other users’ inflammatory comments, rather than cultivating a racist mind of its own will. But if the claim of Tay utilising a combination of an AI team as well as editorial staff, then it’s surprising Microsoft didn’t anticipate Twitter’s infamous tendency for ‘trolling’.
It got to a point where Gizmodo could even compile a list of Tay’s craziest tweets.
Whilst this can be seen as a joke by many, and a disastrous venture into AI that flopped miserably, it’s increasingly difficult to ignore the fact the risk associated with developing AI.
If you think about it, AI was designed to automate processes that would otherwise be tiresome. It’s meant to be efficient, direct and produce accurate, satisfactory results. Tay clearly failed in that department. But what is the single driving factor behind that?
Humans. Human error. That’s why developers are cautious to dip their toes into the AI pool, because what if? What if one day ‘Black Mirror’ became our dystopian norm, because of our actions? Is that just some far-fetched paranoia? Or are there examples of how, potentially, some AI-pocalypse could explode?
The tug-of-war between human advancement and artificial intelligence
One cannot deny that both are linked. We like to think that as we evolve, that means our intelligence goes up.
The evolution of humanity has always relied on technology. The Industrial Revolution might seem basic in principle to us now. However, at the time, the efficiency of so many processes via automation of machines meant increased profits for businesses. It was called the industrial revolution for a reason.
Even if we wander back thousands of years, the invention of the wheel meant a leap forward in transport. Suddenly, you didn’t just have to walk around or doddle along on a horse. You could fix a carriage on the end. If we fast-forward to our time now, we’ve jumped from Facebook and Instagram and YouTube to voice-activated search engines, automated homes, facial recognition on mobile phone cameras, and growing use within the digital marketing world.
For some think tanks, AI can even be the answer to worldwide problems. We live in a world where the population will boom by the billions. How are we going to feed 10 billion people, when there are already food shortages in third world countries?
Will we? If such a powerful AI was built to the point where you could command it to end world poverty, what would it do? Is there a risk of killing all those in poverty, thereby ridding the world of poverty? What if it took charge of food distribution and distributed it evenly, yet so thinly, that poverty became even more widespread?
And whilst the idea of an AI assisting in worldwide problems is honorable, who ultimately programmes it? Controls it? A human, presumably. A human with prejudices; politically aligned; flawed. If an AI is intrinsically absent from bias, surely that virtue is taken away because it needs to be programmed by a human being, and a human being without bias, judgment, political ideas, opinions…is not a human being.
So, it’s a bit of a catch-22. And we’ve already seen what can happen when AI is used for malicious purposes.
Notoriously, manipulation and hacking of AI caused a blackout in Kiev, a suspected cyber-attack. Therefore, it’s likely not paranoia when we say that maybe we should tread carefully over the AI landscape. Oh, we can reap the benefits–and those benefits have proven to be hugely successful!
But the misconception that an AI is a standalone, a machine that thinks for itself…Well, yes, it’s partially true. However, we’ve got to remember that behind the AI–that’s where we stand.
Development? We control it.
The current state of the technology? We control it.
Arguably, that might be the biggest problem.
Can you really translate your brand into an algorithm?
Alas, for all the hype and drama about AI potentially turning evil, some developers aren’t having any of it. Chinese AI guru Andrew Ng, who is a chief scientist for a Chinese search engine as well as a professor at Stanford University, took a drastically different point of view from the likes of Stephen Hawking and Elon Musk.
On the topic of AI and the potential for AI to ‘take over the world’, essentially, Ng said: “There’s also a lot of hype, that AI will create evil robots with super-intelligence. That’s an unnecessary distraction.”
But ultimately, AI boils down to machinations and algorithms, the same way humans are just nuclei and cells.
In regards to just how much of an impact an AI-run business will have on the business, it’s almost impossible to glean accurate data from. The reason is simply that we still require human personnel to programme the initial start-up. And, as we’ve seen with the chatbot Tay, despite apparent human contribution, the self-learning nature of the AI can get a little…Messy.
What about customer perception, too? It’s all well and good if you get a fully functional, AI-controlled business up and running. But do people–do we–inherently trust an AI chatbot as much as we’d trust a human being? If you used a chatbot to help you as a personal shopper, would you feel as comfortable being guided by an AI–who’ll make informed decisions based on the historical data it’s collected from you, and any other data you’ve volunteered–or someone who’s actually got the job title of personal shopper?
Honestly, it’s very individualistic. It’s not something that can be answered generally. Conversely, it’s not something that can be answered on an individual basis, unless you had the time to ask all seven billion people on the planet.
There’s no denying that AI in some areas has given a tremendous boost, business-wise, in some sectors. AI can solve a complex mathematical equation in milliseconds compared to the average mathematician–that’s probably just simple fact.
But we’re living in a time where efficiency is a commodity and time-wasting is a foul. Simultaneously, we shopping experiences to be personalised and effective before we splash our cash.
So, really, if we simplified it down to some core elements, the former supports the quick and clever automation using AI. The latter supports the more personable, individualistic experience offered by a human representative.
Lastly, if we’re in an era where a brand can be identified via its advertising and marketing campaigns, social media interactions, newsworthy stories…Then maybe there is a place for AI. Diplomatically, we’d say there isn’t place for an AI takeover. But there’s certainly place for AI in terms of targeted advertising, and knowing what the consumer is likely to want, rather than spamming all consumers with the same, generic coverage.
But we won’t leave you with that wishy-washy conclusion. Granted, the topic itself is still taking its first baby steps. Wishy-washy is maybe the way of 2018 as we warily watch AI grow. Therefore, instead of jumping the AI gun for your business, how about we take a real-life case instead, and see where things went wrong (and right)?
Case Study: Alibaba
Time for a bit of background.
Singles’ Day is a Chinese event that’s widely celebrated across the country. Despite its fame, it’s not actually an official public holiday. Having said that, it garners the same amount of press as any other holiday would. That’s the scale of how big it is.
And the holiday is, really, what it says on the tin.
Originally intended to be a holiday aimed solely at men, over the years it’s become female-inclusive too. According to the Chinese Public Holidays website, Singles’ Day is “intended to celebrate the lifestyles of bachelors, the holiday is used by some people to bid farewell to their single lives and find a spouse. Many Chinese young people attend dating parties to find a suitable spouse.”
Understandably, before we jump head-first onto the AI bandwagon, other technology had a huge part to play. More and more orders were being placed on mobile phones and tablets and laptops, rather than having individuals take a trip to the actual stores. But, and it can’t be a coincidence, this was also the year Alibaba’s researchers introduced FashionAI.
According to an MIT review, FashionAI couldn’t just recognise items of clothing, but it could also “learn” the tendencies of fashionistas browsing the site. Furthermore, it could assess inventory stock and come up with suitable outfit matches, customised for the user.
Upon assessing this, Abdul Muhammed, Vice-President of rBB communications, said this of AI and its future:
AI leverages what’s best about human behavior – it matches preferences, behavior and shopping patterns with products in a unique way to provide shoppers with a customized shopping experience. Because AI hyper-analyzes data to create digital footprints of each shopper, consumers will see more items they’re interested in and likely to purchase. Increased probability of interest and conversion is not only good for the business, but also beneficial for the consumer who is now receiving a customized shopping experience based on their existing shopping patterns, browser activity and psychographic data from social media, such as what they like, share and post.
Beyond the fashion world
Of course, Alibaba isn’t the only big company to employ AI or even consider employing AI.
What does this mean for us digital marketers, though?
Wading through negative public perceptions, as warned by big boffins Stephen Hawking and Elon Musk, there’s an air of responsibility and sensibility. It may not be a wide assumption to make–that everyone is aware of the risks AI brings. However, when that risk stems from human intervention and the correct programming, then surely it is something we can both regulate and, I hate to be the bearer of bad news, but fear, too?
Yes, individuals can use AI for malicious intent. It’s true! But can we truly let the overwhelming good potential be overshadowed by fear of the bad? Is that truly the society we’ve carved for ourselves?
We cannot acknowledge the darker side to combining AI with behavioural science, which is essentially consumer research. This results in having to monitor people, really. We watch their search history; purchase history; often-clicked items. We even monitor how much time they might spend on a page. All of this data goes into the machine that is the AI.
When we’re living in a world that’s hyper-aware of their rights, and utilising browsers like Tor in order to browse the Internet freely and privately, it’s hard to shift that concern. Because really, the concern raised by whistleblowers like Edward Snowden is perfectly valid. Why shouldn’t you be allowed to browse the Internet innocently, without the knowledge that the MI6 or FBI are monitoring your every click?
Alas, while there aren’t too many examples of fully-functional AI out there in business and digital marketing, there are definitely start-ups. The Alibaba case is a good example. And to be fair, so is Tay the chatbot–despite its disastrous slump.
What we could conclude is that whilst AI is still taking its first steps, it’s also growing rapidly. And if you can utilise this in your business, to create a better user experience and personalise your brand to your consumer, then tackling privacy concerns will be a hurdle. A hurdle you can overcome.
Digital marketing may well be one of the biggest sectors to evolve so quickly. The days of blanket advertising are pretty much shunted into the past. As customers want speed and efficiency, digital marketers strive to meet that demand.
If AI is the way to go this year, then so be it. Just make sure you’re prepared for the upheaval in technology. Don’t do what Tay the chatbot did.
Deliver efficiently and accurately. And maybe, just maybe, you’ll be on the tip of solving the evolving AI dilemma once and for all.