The power of big data and predictive analytics has captured the world’s imagination and added urgency to the rush towards data privacy for many years.

By James Hickman, head of marketing, sales and solutions at Altron Karabina

For instance, a decade ago a US department store aptly named Target made headlines when it revealed it could predict when a mother was pregnant and when she would likely give birth, enabling it to send targeted baby goods coupons to her at the right moment.

There have been too many of these big-ticket announcements to list in a single article, but the point is that big data, and advancements in artificial intelligence (AI) and machine learning (ML), are crucial to modern business strategies. However, what if your entire data stack is outdated, meaning the AI outcomes have little to no value for your business? I’d argue that much of it is, and it’s time to react accordingly before learning lessons the hard way.

Let’s start at the beginning. AI is intrinsically linked to ML, and ML is intrinsically linked to the principle of big data and being able to put those large volumes of data through different algorithms. Ultimately, the machine starts to pick up trends to the point where it can start to predict what is going to happen in the future or provide you with parameters to help make informed decisions.

AI and ML can also be used in a number of different ways to improve customer or user experience, such as an online shopper receiving suggestions that are based on shopper behaviour and broader shopper trends. Obviously, the way these strategies are designed has evolved to keep pace with changing privacy laws around the world.

This is all good and well, but it is potentially irrelevant now because anything pre-pandemic is not really relevant to the current environment. It’s probably best to explain this by way of example. Imagine a business development team trying to predict where to put the next petrol station. You can’t use the data sets around traffic patterns, because pre-pandemic patterns were vastly different from what they are today.

What about during the pandemic? We had the hard lockdown where no one was driving, then we had sporadic ups and downs depending on the level of lockdown, and now as we emerge from the fourth wave with signs we are headed for the end of the State of Disaster, some people acting are in a fairly “normal” manner, albeit in some instances leaning more towards a hybrid-working model.

So where does this leave us? We almost find ourselves in a world where pre-pandemic data is arguably irrelevant and data from during the pandemic is overly dependent on point-in-time scenarios. Every business should ask: Can you actually get accurate information relevant to today from the data that you have collected?

Obviously, there is data that falls out of this category that wouldn’t necessarily be as intrinsically affected, but even if we were only to look at consumer trends, one could argue that there aren’t that many usable data sets – certainly to the degree that there was pre-pandemic.

This calls for artificial intelligence to be more intelligent because you have to write into the algorithms learnings relevant for our pre-pandemic, mid-pandemic, and even post-pandemic datasets, at a whole new level of detail or else you will just end up with answers to the wrong questions.

Businesses have spent millions of rands collecting, sorting, storing and processing data without necessarily appreciating that the pandemic shifted our world and consumer behaviour so much that the way AI and ML are being applied to this data to make predictions may well be irrelevant in 2022 and beyond.

Artificial intelligence was a megatrend leading up to and during the pandemic, but it is really now a megatrend that needs to redefine itself or redefine how it is utilised by businesses and their partners to reap its intended benefit.