In fact, he says, the technology is so promising, companies can indulge in "moonshot-type" projects. In other words, the return on investment in AI is so great and so certain that companies can’t afford not to be daring in their cognitive computing initiatives.

Selle elaborated on his views in a conversation with SupplyChainBrain editors at the recent North American Supply Chain Executive Summit, held in Chicago.

SCB: Joe, you make no bones about: you’re convinced that the potential for artificial intelligence and cognitive computing is extremely broad throughout business. So make the case; what leads you to that conclusion?

Selle: There really is broad potential. At this point in the evolution of our thinking around business, I would say we’re at a pivot point. We’ve been through descriptive analytics and through predictive and prescriptive analytics, and now we have a step change in front of us, which is going to be moving into the cognitive era.

As you move from the more traditional analytics — though they are still advanced — into cognitive computing, you have a big step change. You have a lot more value at stake. You have the option to create moonshot-type projects in your companies. It’s very broadly applicable.

We’re here [at the Summit] to talk mostly about supply chain, but you can apply this kind of approach to any operational challenge, whether it’s a finance process, sales, marketing, you name it. By the way, as we get more into integrated business planning, all of these functions weave together to impact the supply chain anyway, so they are all part of an extended supply chain universe, if you will.

SCB: So the applicability will be extensive as well as expansive, in you view. But let’s drill down more into supply chain operations. You say they can be completely reinvented. How so?

Selle: I think what leads me to that conclusion is that the approaches we’re taking with cognitive computing are very new. They really are about reinventing the process. Plan, source, make, deliver are traditional supply chain processes. There’s return, there’s procurement — there are related customer-service-type processes — and when you take the cognitive principles of a system that can understand, that can reason, that can learn and that can interact almost as naturally as we’re talking — not quite yet, but it will at some point — URLI — understand, reason, learn, interact — these are new tools we can bring to business challenges. And when you have systems that can do that, you can truly reinvent the business process.

You can drastically shrink the business process, you can take any routinized behaviors that your people are doing, take those down and you can shrink them in time, you can improve their accuracy. The people are still making the final judgement call, but now they have a digital assistant sitting next to them, on their iPhone, on their tablet, on their desktop computer. It’s very easy to access it, and that system understands what your job is. You’ve trained it. It will help you make better decisions. So you can take that and reinvent your business process.

SCB: You make a compelling argument about access, but what about to quality of data. That’s always been a challenge. Going forward, how are companies raising that concern going to address that issue?

Selle: They would be right. Data is a huge challenge. It’s notorious for anybody in any function, but particularly in supply chain where you have extended partnerships. You might have contract manufacturers, logistics partners, you’ve got people in different countries all running different systems — these bits of data are coming at you in very different formats and at different levels of quality — so you have to address that if you are to build a cognitive system.

But the way that people who are successfully building these systems have addressed it is to take small areas first. They’ve taken the domain of logistics or order management or something that is somehow confined. They’ve worked on data quality for that particular challenge, they’ve applied cognitive computing to address it, and they enjoy the business benefits as the processes improve. That’s sort of iteration number one.

Iteration two is that you make more of a foundational investment into a data lake or data platform. Now you have a place in the cloud so all the players can get at it, and you have formats, and meta data standards where all the data can come in. And once it’s in the lake or in the enterprise data platform, it adheres to certain characteristics, in terms of the way it’s structured. You can trust it. And there are owners of the data so, there’s a data governance aspect.

So you go from a single enterprise project where you work on the data for that project to a cloud-based solution where it’s now become a data lake or data platform, and then various business users can get into that platform. They can access the data for whatever their purpose is. Now you’ve got people coming in who are just business users, not data scientists, not operations PhDs, just normal business users. They can go in and access data in a way like never before. That’s iteration 2.0 — or actually it may be 3.0.

SCB: Clearly, many folks want to hear about the ease of use, but what about those folks who don’t use the technology but who are empowered to authorize the investment in it? Give them an example of how AI is beneficially affecting the supply chain.

Selle: I would say that those executives who are funding these initiatives are doing so because the business benefits that are promised are much bigger than any they’ve ever seen from any other project. The ROI is quite high for these cognitive investments, and I can give example.

Let’s say you have a team of lawyers that are reading supplier contracts. They may have thousands of contracts under management. If they’re like IBM, they have hundreds of thousands of contracts with suppliers, for buying a lot of things: buying labor, janitorial services, direct materials to put into computers, logistics services, you name it — it’s all governed by these contracts. Now you have some sort of contractual change that you need to implement across the entire field of suppliers. How do you do that? How do you know if you’re in or out of compliance? The lawyers or contract specialists have to read all of the contracts and make a determination: these comply, these don’t.

What if your digital assistant, named Watson in our case, were to read those contracts and give you back a list of the ones that are most out of compliance with your standard terms and conditions? You’ve taken a team of specialists, maybe several dozen, working continuously across some sort of period when you’re, say, bringing an acquisition into the company, and you’re giving them a tool that will make their work go possibly 70 or 80 percent faster. That’s a pretty compelling example.

SCB: Applicability beyond that?

Selle: We all have contracts. You can apply that same logic to other regulations that are hitting your business — environmental regulations, financial regulations, revenue recognition regulations. If things are changing, and they’re different in every country, how do you know if you’re compliant? The old-fashioned way is to make sure you’re reading all of these regulations that authorities are promulgating, but why not have a digital assistant read all that for you and tell when you have a potential problem?

SCB: So, if someone wants to explore the value of AI, what would be your advice to them about next steps? How should their company proceed?

Selle: It’s very important to have a visioning phase of any of these projects. The visioning phase is where you come up with new ideas to reinvent the process. We have a cognitive garage process at IBM which we employ across all functional groups. We start by looking at their business processes, understanding where the persistent challenges are and then looking at how to apply cognitive capabilities to get to a next-level solution. So we start with a one-business, one-challenge [methodology] at a time. We come up with a journey map for cognitive evolution.

We make sure we understand what the first piece is very clearly, and that first piece has to be small, even though you've spent a lot of effort thinking big. So my advice is to think big and then start small and then scale it quickly, if you think you have something that works well.

Click here to view the video 

Resource Link:
IBM