Our industry just loves a buzzword or a theme du jour. We’ve recently, and hopefully finally, moved on from every second article being about Social and Viral. Now we’re onto Big Data and Content and, in the last few months, they’ve been joined on the circuit by the latest buzzwords, Artificial Intelligence (AI) and virtual personal assistants (VPAs).
Many of these themes never really live up to the hype the column inches suggested. For example, both the internet and social media were introduced as the end of advertising, yet ironically Facebook now sells itself primarily as a paid-for advertising channel. There is evidence to suggest that brands are more important in an online world not less (a key reason being that unlike with shops you don’t tend to inadvertently visit a website you’re not looking for). Thankfully viral as an objective has also faded largely because only creative outliers, like Cadbury Gorilla or Volvo Epic Splits, managed to achieve much virality or interaction. Furthermore the viral success these brands did achieve didn’t end up being very easy to replicate.
So, is our latest buzzword, Artificial Intelligence or AI, going to be a big deal or is it something that maybe we could all ignore for a year or two until it goes away?
Predicting the future is difficult; predicting the future of technology even more so. Indeed as UK futurist and comedian Warren Ellis wisely states, if you want to avoid looking like an idiot it’s best to avoid the business of prediction altogether.
One doesn’t have to look very far in the world of technology prediction to see evidence of idiocy. Take Sir William Preece’s terrible prediction on the uptake of phones in England; “The Americans have need for the telephone but we do not. We have plenty of messenger boys”. Perhaps we can give him a pass as he said this in 1878 and we didn’t know very much back then but even in the last few decades we’ve had such gems as “there is no reason anyone would want a computer in their home” and “the truth is no online database will replace your daily newspaper, and no CD-ROM can take the place of a competent teacher”. And it’s important to point out that both these guys were engineers and one even had a PHD, so they were certainly no fools.
One of the problems with technology predictions is that they tend to be done by engineering types; people with lots of technology knowledge but not so much human behaviour knowledge. But human behaviour knowledge is a rather important input to use when assessing the likelihood of people embracing a technology or not.
What the myriad of experts in the field of human behaviour and decision making tend to all agree on, is the structures and default ‘settings’ of the human brain have remained largely unchanged over hundreds of thousands of years. They are optimised for the lives we used to live in small, homogeneous, hunter-gatherer societies. Viewing human behaviour through this lens makes it easy to see why:
- we like easy, fast decisions (expends less energy when energy was scarce);
- we prefer group behaviour (all fight or all flight was better for survival);
- we distrust new people (they could be dangerous);
- we’re terrible at saving (it was unnecessary);
- and we over eat (we didn’t know when the next meal was coming).
The history of technology uptake tends to reflect typical hunter-gatherer human behaviour rather well. Technology that fits with a brain wanting the most cognitively easy solution tends to be successful and that which doesn’t tends to not be.
Apple is a very successful company. Their products are intuitive and very easy to use, so much so that they require no instruction manual and even young children learn to operate them very quickly. In contrast 3D TVs and Google glasses have not achieved mass uptake. They are both annoyingly tricky to use. Similarly, looking back in history at successful technology we can see that phones are easier than messenger boys, MP3 easier than CDs and Google Search beats the Yellow Pages for effortless results, hands down.
So a technology that removes the need for typing and makes interacting with all the information on the Internet as simple as talking to your very own Mr. Spock, will be very appealing for our efficient, effort-minimisation-seeking brains. As will a technology that knows enough about your behaviour to start pre-empting your need to ask it questions. For example your Spock VPA will know that because you missed your outbound flight you will need new flights for your next leg, new hotel bookings and it will cancel and rebook the upcoming meetings you’ll be missing. Or, when looking for a new job, your VPA will know what kind of companies you like and ones that are likely to like you and make recommendations accordingly. Furthermore, if instructed, it could liaise with said companies VPAs and help whittle the list of options down even further.
So, it seems reasonable to conclude that AI & VPAs are almost certainly more substance than snake oil and will have a rather large influence on our lives, both personally and professionally – and it isn’t all that far off.
Looking back over the last few years the growth in AI achievements is pretty impressive. In 2011 IBM used AI to beat the world champions at the TV game show ‘Jeopardy’ (which from a computer programming point of view is much harder than playing chess).
Google used AI to create neural networks that can recognise faces and animals. They also used AI to map every single French business, household and street number (a task, by the way, that was completed in less than an hour).
Skype translate can now almost translate in real time and in 2015 Samsung released TVs that now have voice recognition software in them and can understand and capture conversations in the rooms that they’re in.
We know that data, algorithms and computer processing are all increasing at exponential rates. Furthermore, the data is increasingly interconnected which dramatically increases its usefulness and accuracy.
What does all this mean for marketing? In some respects it’s just the realisation of the big over-promise of digital marketing for the last decade, i.e. more relevant messages. In an interconnected AI world, marketing messages will actually be more relevant (no more annoying retargeted messages for something you bought months ago or ads from categories you don’t buy).
The biggest advances are likely to be in direct response and retail marketing. Today when we want to create an immediate response we tend to deliver incentives and discounts en masse. Many offers are served to customers who would have happily paid full price. In an AI world we will know who is in market and who is out of market and who might be able to be incentivised into market. We’ll also know what type of incentive is more likely to be more influential for which people. We’ll be able to genuinely add value, which will result both in more profit (less discounting) and happier customers – a win-win for everyone.
The largest implications are likely to be for low interest categories. With an empowered VPA able to conduct shopping in categories consumers have little interest in, ‘low interest’ will easily become ‘no interest’. This is great for consumers but not so good for companies; especially those that aren’t number one or two in category and to which most consumers currently default.
New entrants in these categories will need to find ways to influence a VPA that is making the decision. One way will be to be cheaper so these categories are likely to become even more price sensitive and they will likely become categories where the marketing is largely machines talking to machines.
For the advertising industry (and also the general population) perhaps the best evolution will be in brand communication. Its death has has been predicted many times in the last decade; often by those in the digital and social media worlds. And in a world where VPAs can analyse reviews and choose products on a consumer’s behalf, some people are probably going to predict its demise yet again.
However, advertising is highly likely to become even more important than it is now. Media owners will use ‘quality scores’ to give preferential treatment to the more liked, more viewed or more spoken about brands. The preferred treatment will mean better pricing (as Google and Facebook are already doing) and probably also better media placements.
The best thing about scores based on ad likeability data, ad interaction data and product reviews is that the kinds of ads that float to the top are those like Volvo Epic Splits, Cadbury Gorilla, the John Lewis Christmas ads and almost anything by Nike. What is common to almost all is that they are much more entertainment than information – implicit messaging rather than explicit. When the economic benefit of this type of communication becomes more apparent than it is currently, the incentive to make them will increase. So great creative ideas and great ads will almost certainly be more in demand, not less. That benefits anyone watching TV or AV content regardless of what kind of screen it is on.
Furthermore, whilst VPAs obviously won’t be affected by brands some of the data they’ll be analysing certainly will be. Product reviews are heavily influenced by how a user feels about a brand. Currently not everyone reads reviews pre-purchase, not least of which because it’s time consuming. In a VPA world many more people will be influenced by reviews because their VPA will ‘read’ the reviews for them. The stronger, more liked brands will get better reviews, which will then further boost the brand’s quality scores.
In summary, I can’t tell you when machines will be as smart as us or when they might become fully autonomous. Although I rather like the filter that software architect Brad Templeton has suggested; “you’ll know when a robot is truly autonomous when you instruct it to do some work for you and it decides to go to the beach instead”. But, I can suggest that unlike many of the new shiny things with which our industry gets obsessed, AI and VPAs are going to be a very big deal. The impact will vary by category but marketers might want to get comfortable with more esoteric, implicit, entertainment based brand communication; much more analytical retail that is far less based on discounts; and, if you’re in a low interest category perhaps you should boost your preference and market share now before a VPA locks in default settings that you may not like.