Morality vs. ROI: AI and Algorithms Aren’t Always the Answer
There is a perception in the digital marketing world that if you can automate something, you should. There are sophisticated algorithms and automation software that can take over many roles within a company, carrying out tasks and processes so people don’t have to, but removing the human element isn’t always a good idea.
In his recent talk at Spark.me 2017 Rory Sutherland of Oglivy looked at just how nefarious and essentially practical algorithmic marketing can be. Looking at automated and algorithmic advertising, Sutherland highlights how this sophisticated technology’s focus is almost always to maximise returns and this means targeting people who may not want to be targeted and finding their weaknesses. Algorithms are driven by results and the idea of reputational damage, offense or even physical harm to an individual simply isn’t part of their makeup. To describe it as immoral is a little unfair, computers and technology do not have the ability to “care” as such but the point is, relying solely on algorithms is a dangerous approach for any business with a reputation to manage.
Ethical Marketing: Managing and Moderating Algorithms
As marketers the focus on ROI is always high but taking it too far can be dangerous and cause more damage than positive return. No business wants to be responsible for bringing gambling addicts back to an online casino for example, or advertising alcohol to alcoholics, yet with an algorithm-based marketing campaign it is highly likely this is happening.
Many marketers don’t have the awareness of how algorithms work, they simply see the potential return and run with it. There is no morality or peer pressure for an advert, it simply performs in the best way it can, targeting its prime customers.
Automation Vs. Humans
It’s not at all a surprise or a secret that automation and algorithms are quicker and are ideal for solving problems that involve a huge amount of data and following patterns. They are perfect for data entry and repetitive tasks whereas humans are more prone to error and don’t have the same abilities to evaluate patterns in big data. These are areas where it makes natural sense to automate and make use of algorithms.
AI and algorithms have no ability to deal with grey areas or ambiguity and this is where the risk comes in for any business. Context and nuances are not a consideration and they cannot make the right judgement call in many instances. To go back to Sutherland’s point, no business really wants to advertise alcohol to alcoholics, yes, the return would be high, but the reputational impact would be far from positive.
To use a further example from Facebook, whose algorithms repeatedly get them into trouble, there have been many examples of their “Year in Review” or “On This Day” features bringing up images and memories that cause real pain to its users, simply because the image was popular at the time. Algorithms have no understanding of good or bad, they can simply use the data in front of them to deliver what the feature in question is designed to offer.
More of the economy is becoming digitised and automated but the larger companies are in fact employing more people to oversee this, with data being pruned, cleansed and sorted before it is even close to being ready for analysis by algorithms. Humans are a necessity to ensure that big data can be used efficiently, and the ROI delivered is ethical and beneficial to business, not just designed to get the highest return, regardless of the damage and upset it may cause.
