Computer-based algorithms are capable of making accurate calculations at a scale we can barely even conceptualize, never mind replicate.
As such, they are an incredibly powerful tool, to the extent that they manage a lot of our quotidian tasks for us already.
Much of what we do daily in digital marketing is rule-based, so it therefore stands to reason that it can be automated in this way too.
We are now moving into the realm of sentient artificial intelligence, capable of learning from surrounding stimuli and making ever more informed decisions.
Great strides have been made by technology companies in all areas, from the mundane to the creative, to provide better results by incorporating machine learning systems.
Not all of this is new; much of it has been available or in development for many years now.
However, the addition of the concept of ‘deep learning’ has expanded what is possible and has also increased the likelihood that some jobs may no longer require fulfillment by people.
The finite intellectual, perceptual apparatus at our disposal require the aid of machines if we are to move into an age of unprecedented progress.
We, as digital marketers, have inevitably been discussing this as either an opportunity or a threat. However, the topic is complex, and unfortunately, complexity and comprehension rarely go hand in hand.
If we really want to understand how and when artificial intelligence will come to play a central role in digital marketing, we would do well to begin with a deeper understanding of the component parts of what we do.
Within this article, I will look at the historical precedent for what we are experiencing, which will shed some light on what we should expect. Furthermore, I will look at a new categorization of tasks – one which might allow us to understand how best to avail of artificial intelligence.
The shock of the new
History is cyclical; patterns play out repeatedly over time. The pace of technological change may be dizzying, but the pace at which people change could generously be described as glacial.
As such, the shock of the new remains as potent as ever. We need only look to the scaremongering around the launch of the bicycle, more than a century ago, to see evidence of this.
In Cycling: The Craze of the Hour, written in the 1890’s (just as the bicycle was assuming mass popularity), the author warns:
“Directly you are in motion you will feel quite helpless, and experience a sensation of being run away with, and it will seem as if the machine were trying to throw you off.”
Sound familiar? There are echoes here of the present-day warnings that sentient machines will take over, that they will overpower us.
By now, we should really be kicking back with our feet up while the robots take care of the drudgery for us.
And yet, here we are. We work more hours than ever; we constantly claim to be ‘busy’; we somehow manage to fill every waking hour with ‘work’ of some sort.
What are the machines waiting for?
One potential answer reason some jobs aren’t automated already relates to the ethical questions resident in the widespread use of unsupervised artificial intelligence. This has been described as AI’s dark secret (again, we should remember the fear of the humble bicycle a century or so ago), as the system provides little transparency into how its conclusions were reached.
However, much of the ethical discussion that surrounds this issue resides in industries that are, put simply, more important than ours.
The repercussions of an inaccurate AI doctor or a faulty driverless car are evident. The specter of sub-optimal ad copy doesn’t send quite the same chill down one’s spine.
That is important to note, as CMOs will be willing to take the plunge with their advertising dollars, knowing that there is much to gain, but not so much to lose. The tasks of defining customer segments or deciding on channel attribution do not attract much in the way of sentimentality; if the machine can do it better and cheaper, the machine will get the job.
Another, equally plausible, explanation lies in our own perception of what it means to work. In ancient Rome, for example, paid work was seen as demeaning. In the eighteenth century, great significance was given to leisure time, spent discussing intellectual questions in cafés or salons.
Through the twentieth and twenty-first centuries, we have come to link profession inextricably with both pride and profit.
We control this system, let’s not forget, so we will keep ourselves in work if it is deemed to be for the greater good. Technology is nothing more than an extension of our being; it does not necessarily replace us.
Moreover, the idea of a 10 or 15 hour work week opens up all manner of questions for the wider populace, and for governments too. Many people are worried for their jobs, not excited by the prospect of increased leisure time.
The introduction of a universal basic income, for example, seems sensible and may erode some of those concerns over time, although its implementation would be far from seamless.
Nonetheless, it would still not satiate the greed that feeds so many careers.
A new concept of what a profession is will likely be required if we are to vault this hurdle. The notion of a vocational career will be replaced by specific skills that blend best with the capabilities of technology.
As such, we should first decide on how that relationship might function.
Which tasks will be taken over by automation first?
In a recent ClickZ Intelligence report, we saw a typical categorization of trending tasks in 2017, along with the significance of each for marketers:
At each of those stages, it is possible to make a case either for or against the widespread use of automation. What is undeniable is that automation of certain tasks, like pulling search volumes or calculating site-wide metrics, saves time and allows us to focus on more interesting work. This has been happening for some time already.
This segmentation makes sense to us, too. We cut things up into bite-sized chunks, so they are easier to understand – and to sell. But that doesn’t mean that it is the best – or only – solution.
We could cut across all of these tasks with a categorization of the following:
Sensory: Interpreting results; understanding a client’s business and applying this knowledge.
Interpersonal: Relaying the findings from a campaign to different teams and deciding on a course of action.
Mental: Idea generation for campaigns; assessing competitor content strategy.
Mechanical: Building out account structures; keyword research; meta data creation.
Now if we combine the two lists, taking tasks like content marketing and seeing that this requires sensory, interpersonal, mental, and mechanical input, the scale of the challenge becomes apparent.
So if our fear is that automation will take over “everything”, we should remember not only that such complexity in our work exists, but also appreciate that it is in applying – and integrating – all of these skills that we are successful at what we do.
The machines are getting better at each of the requisite areas, but it is in stitching these together into a cohesive, contextual plan that we get the best results. A machine that is capable of the sensory, the interpersonal, the mental, and the mechanical all at once might achieve this.
However, we should bear in mind that the way a machine (even a ‘sentient’ one) perceives the world is different to the functioning of the human brain.
If we can assume a more sophisticated understanding of those differences, we can arrive at a conclusion that plays to the strengths and weaknesses of both parties.
At MIT, this is referred to as ‘extended intelligence’, as machines allow us to go beyond what our limited faculties provide. This is a healthier and more helpful view of what these exciting technological developments could mean for all industries.
What will this mean for marketers?
Marketing is about communicating with people. Therefore, the input of each campaign will likely remain human, and the output will, of necessity, involve human interaction. It is in the processes of marketing that we will see a fundamental shift.
Google has stated that it will not go “100% machine learning” with its search algorithms, primarily because it becomes impossible to debug a machine if you don’t understand its processes. Human activity will remain essential to Google, as it will to most other companies.
That said, we will need to be more flexible than we currently are and the notion of a long-term career may vanish. Nonetheless, we have come through much bigger shifts than this over the past few centuries and dystopian visions of a computer-dominated society are hardly new.
Those with the means and the money at their disposal will decide our course.
So what matters to this select, exalted few? Well, automation will decrease costs and increase returns. A marketer’s dream, in effect. That should provide us with some pretty useful clues.
These are undoubtedly fascinating times, but the pattern played out over the next ten years will closely resemble one we have seen before.
If one thing is for certain, it is that people will never be satisfied with their lot, and history also tells us that these increased returns will be reinvested. We will find a way to keep ourselves busy at our work.