An AI strategy is important when you don’t know what you’re doing. Might AI uncover that great tactic or logistical issue? It might, but then again it might not. A lot of AI solutions are black boxes – you don’t quite know what goes into them or why it comes out the way it does. That might give you a great short term answer that blows up long term (or when an unforeseen event like a pandemic comes along). Most retailers never took the initial learning step – here’s what drives my business – and AI is not what they need.

I believe you have to be in it completely to see the real benefits of AI. That means having a solid infrastructure, clean and accurate data, and personnel to support the investment. As mentioned in earlier articles, I think the investment in AI is moving faster due to the pandemic so that companies can conduct business in an ever changing landscape. The companies must also be able to show the worth of the investment otherwise I believe buy-in will be a hard sell.

The most important piece of AI is really thinking through the problem it will help solve. What is the practical application of it? Many times AI is included for the wrong reasons and it ends up being a solution looking for a problem to solve.

If a company cannot articulate what they are using AI for or what they are trying to achieve, then they deserve to fail. That would be the same for any software they purchase. Why is AI seen as a mysterious thing that when deployed by a company will solve all of its problems? Let’s get real here, AI is a tool like any other — if you do not have a direction the tool is not going to help you find it.

First of all, recognizing data and analytics as an enterprise asset and competitive necessity is essential – perhaps obvious, but far too many have yet to embrace this reality.

In my experience, successful prioritization of AI projects takes into consideration a blend (value v. effort) of tackling low-hanging fruit (aka quick wins) to sustain momentum, incremental improvement of enterprise business processes to ensure cross-LOB engagement and fleshing out data governance improvement opportunities, and a few “moon shots” to test new hypotheses and drive innovation.

AI is a holistic approach to problem solving. Think about a self-driving car. The AI involved cannot separately consider the operation of the car in technical silos. It can’t consider the operation of the car without anticipating the driver’s reactions and behavior. It can’t safely go on the road without anticipating every other car and its behavior — and more. Not one action can be programmed in isolation.

It is the same for retail. AI is not an analytical approach. It is a systems approach. The difference between information systems and technology and AI is that proper AI sees everything and understands how EVERYTHING interacts and affects everything else.

How should an executive prioritize AI strategy? It depends on the time frames. If the executive’s focus is five to 10 years for the company, prioritize it #1, because it will take care of everything else. If the executive’s focus is next year, don’t even start.

AI cuts through the noise in today’s retailing environment. Developing an artificial intelligence strategy is essential to every business, large or small. There are plenty of ways to get started with existing platforms. Google Analytics uses machine learning algorithms and is easy to access. Facebook and Instagram offer features enhanced with machine learning algorithms. For big companies, there are many new platforms available to run pilot programs. For example, I am excited to see Mastercard’s Shop Anywhere pilot with Circle K, Dunkin’ and White Castle. The answer is to make AI a priority and find the right platform to get started. Period.

AI can best be used as an enabling technology to help retail businesses accelerate previous people-centric analysis and resulting decision-making. Retail executives should prioritize AI as part of their continuing digital transformation efforts, not in lieu of them. And these should be balanced against all of the other efforts required to run and grow the business – rather than overtake them.

AI based prediction, which is a judgement of an individual customer based upon a myriad of algorithms interacting to find the “correct” answer as determined by an individual programmer, is ripe for bias. At this point in time, the retailers who have invested time in grasping AI are focused on the rudimentary steps of a one-size-fits-all solution. Personalized customer experiences, removing friction, etc. is not intelligence, rather a mapping of behaviors translated by non-intelligent systems of algorithms. Those retailers crossing into the realm of prediction/judgment of an individual human customer outcome on the behest of their internal AI models must prepare themselves for the unintended consequences of the inherent bias of AI models designed as a one-size-fits-all solution. AI diversity is not an option if retailers are sincere. “Playing” in AI has the ability to create hidden landmines, waiting to be exposed in heretofore unimagined human scenarios. Retail executives, to truly lead in AI, should take a deep dive into the documented groundswell of the across-the-board AI gender bias hidden in plain sight. In time,… Read more »

AI can provide new options for solving old (largely unsolvable) problems. As others have pointed out though, the right path is not to take AI and find a problem worth solving, rather it’s revisiting existing roadblocks and determining if new tools can provide new approaches. Without a well defined use case, AI will be no more successful than any other prior attempt.