The hype surrounding AI is real. You know it, I know it, we all know it. So what exactly is the problem when it comes to adoption? Do certain misconceptions of what AI can and can’t do factor into all of this?
As the consumer evolves and legacy retailers scramble to ride the high from last year’s positive holiday season, investments in AI continue to dominate conversations. In fact, global retail spending on artificial intelligence is expected to grow to $7.3 billion per year by 2022. This is a HUGE increase from the estimated $2 billion spend on AI in 2018.
The expected growth in AI comes as no surprise, as many believe in the promising future this technology holds:
"In the sphere of business, AI is poised have a transformational impact, on the scale of earlier general-purpose technologies. Although it is already in use in thousands of companies around the world, most big opportunities have not yet been tapped. The effects of AI will be magnified in the coming decade, as manufacturing, retailing, transportation, finance, health care, law, advertising, insurance, entertainment, education, and virtually every other industry transform their core processes and business models to take advantage of machine learning." - The Harvard Business Review
However, a few misconceptions still exist, leaving others terrified of its impact or hesitant to adopt. A recent report made it clear there is still a huge divide between leaders and laggards in the application of AI across and within various sectors (not just retail):
So, what gives?
First off, gaining a clear understanding of what’s necessary to run an effective AI system is critical. Seeing as AI technology is still in its early stages and continues to evolve, having a strong understanding of its boundaries and abilities will be instrumental to help bridge the gap between adopters and laggards over time.
The Volume of Data
"We all know that artificial intelligence needs data to learn about the world, but we often overlook how much data is involved. These systems don’t just require more information than humans to understand concepts or recognize features, they require hundreds of thousands times more." - The Verge
Lots and LOTS of data is necessary for the successful use and implementation of AI.
It’s a must. An article from The Economist couldn’t have said it any better: the world’s most valuable resource is no longer oil, but data.
Luckily for most retailers, data isn’t the problem. Retailers have loads of data about their customers. An overwhelming amount of data actually, much more than they can handle.
POS data, social data, e-commerce data, you name it.
The problem is a retailer’s data is often spread across various systems or cluttered in Excel spreadsheets, and often divided among multiple departments. This is a major pain point for retailers as it makes it increasingly difficult to paint a clear picture of their consumer and understand true product demand across thousands of store locations.
This is exactly where AI fits in.
With all the existing data siloed across systems, machine learning algorithms can help retailers leverage all the information they have to depict a clear picture of the consumer and future demand across channels.
All the consumer behavioral data necessary to feed machine learning and AI technologies is exactly how Amazon became the Goliath it is today. They’ve been collecting consumer data since their inception and use this to drive every business decision they make.
Why does AI technology need so much data?
The answer is rather simple:
I came across a great example, to put this into perspective a bit, of how much data AI systems need to train. Apoorv Saxena, the lead product manager at Google, compares the abilities of humans and AI systems to distinguish an image of a cat and a dog:
“Humans do not need to look at 40,000 images of cats to identify a cat. A human child can look at two cats and figure out what a cat and a dog is — and to distinguish between them. So today’s AI systems are nowhere close to replicating how the human mind learns. That will be a challenge for the foreseeable future.” – Wharton, University of Pennsylvania
Understanding the need for large data sets is critical if you’re looking to leverage AI in a valuable (aka profitable) way.
Training the Machine
While a massive volume of data is necessary to effectively fuel any AI engine, the quality of that data is also really important.
Simply put, good data in means good data out.
The main thing to understand here is that data is the oil needed to train an AI model. Yes, you heard me. It has to be trained, which is the fundamental concept behind this technology.
Ah, now the name makes sense, doesn’t it? Essentially, we need massive amounts of good, quality data.
The better-quality data you have going into the algorithm, the more accurate the predictions coming out of it. Makes sense considering AI isn’t just some magic crystal ball accurately predicting the future (although, it kind of seems like it is sometimes).
For retailers, at least two years’ worth of good, historical data is needed to get an AI system up and running.
This, however, is just the beginning.
Training the model is an ongoing process that requires a constant flow of data (large amounts and good quality) to continue making improvements to your AI system over time. How long you’ve had your AI system in place doesn’t drive improvement, the data coming in every year does:
“[T]he ongoing value of data usually comes from the actions you take in your day-to-day business — the new data you accrue each day. New data allows you to operate your prediction machine after it is trained. It also enables you to improve your prediction machine through learning. While 10 years of data on past yogurt sales is valuable for training an AI model to predict future yogurt sales, the actual predictions used to manage the supply chain require operational data on an ongoing basis. And this is the important point for today’s incumbent companies.” – Harvard Business Review
I know I keep reiterating this point (but I’ll say it again)—the amount of data and the quality of data flowing into the algorithm go hand-in-hand into training an AI system effectively.
Explaining why an algorithm came to conclusion X, Y or Z is also another large factor that may play into slower adoption rates.
"Another major challenge is understanding how artificial intelligence reaches its conclusions in the first place. Neural networks are usually inscrutable to observers. Although we know how they’re put together and the information that goes in them, the reasons why they come to certain decisions usually goes unexplained." - The Verge
It can be difficult to buy into the results AI brings without a clear explanation of why it came to a certain conclusion. This is especially true for anyone skeptical of AI in the first place.
It’s even truer considering how costly an AI investment can be, in addition to how costly it could be if the predictions are wrong. People have to take responsibility for the conclusions machines come up with, so having access to ‘the why’ is a completely valid concern.
The power behind AI technologies can be sort of a double-edged sword – they excel by developing this whole new way of seeing things, yet their outputs may be incomprehensible to the way humans see things.
So you can see how the “why” sometimes lacks transparency, which is what most refer to as the “black box” problem in AI.
However, there are AI tools that do provide such ‘explainability.’ At Celect, for instance, we can provide the “why” to back up the recommendations we make to retailers optimizing their inventory through our patented technology. Answering questions like where the right allocation is for a particular product at a store location or how well a new product introduction will fare at another store location (will it cannibalize my existing assortment?) are good examples of such explanations.
The ability to answer these questions and see the potential effects inventory changes could have on your bottom line is crucial.
Be a Trailblazer
Despite the limitations of this evolving technology, the opportunities it presents for retailers are undeniable:
"Although it is hard to predict exactly which companies will dominate in the new environment, a general principle is clear: The most nimble and adaptable companies and executives will thrive. Organizations that can rapidly sense and respond to opportunities will seize the advantage in the AI-enabled landscape. So the successful strategy is to be willing to experiment and learn quickly. If managers aren’t ramping up experiments in the area of machine learning, they aren’t doing their job. Over the next decade, AI won’t replace managers, but managers who use AI will replace those who don’t." - The Harvard Business Review
Using AI to turn data into actionable insights is the future for many industries and is especially true within retail. It's time to gain an understanding of how you can leverage the evolving technology to be at the forefront of this transformative period. Data-driven decisions will be key to successfully overcome the challenges permeating outdated business models across retail industries and beyond.