AI/Machine Learning for Supply Chain: Into the Future—Part Three
New System Adoption Lifecycle Required for AI/ML-based Systems
on Nov 9, 2021
AI/ML requires a reimagining of the system development and adoption lifecycle. We discuss the move to a more agile approach, potential use of AI/ML for data cleansing, and the new skillsets and changing roles and responsibilities required.
In Part Two of this series, we examined the potential use of AI/ML in enabling continuous planning and execution, and as a foundation for autonomous supply chains. Here in the third and final installment, we discuss how AI/ML systems require a different development lifecycle and skillsets.
A New Systems Lifecycle
One other important area that needs change now is how we adopt new systems capabilities. AI/machine learning requires a new model (see side bar) that takes advantage of the learning environment/capabilities of the technology.
Companies are looking for more agile approaches that allow users to rapidly engage and use the solution. Agile development/implementation is especially helpful in the dynamic world where change is upon us and we need to react quickly.
Agile lets users learn about and use the new technology in bite-sized, therefore, lower risk sprints.
So, what does that lifecycle look like? Figure 1 below proposes such a lifecycle.
Discovery is something we often turn to analytics for. The old systems lifecycle approach leaves analytics to the end of the process, after the rest of the transactional or planning system has been built and a data warehouse installed to support this reporting. That’s way out in the distance! One of the gems of a machine-learning lifecycle approach is that users can get a glimpse of those new insights as the capabilities are being created—not way down the lifecycle.
Another bright spot in the AI/ML development and implementation area is using AI/ML for data cleansing. This arduous, tedious, and time-consuming task of data acquisition and cleansing is reduced. Rather than purchasing, installing, populating and cleansing data as a precursor to even getting an opportunity to engage and use technology, we get engagement much earlier in the process.
Figure 1: System Adoption Process/ Lifecycle
Changing Roles and Responsibilities
Another area that should be addressed is the roles and responsibilities of both supply chain and IT. So much is on the shoulders of supply chain pros today that they should be given the due respect and support appropriate to all they know and all they do. Many of them have deep economic, technological, statistical/mathematical, and business skills. We, therefore, call these professionals Supply Chain Scientists.
With the advent of the so-called data scientist, the hunting, analyzing, and, often, categorizing of data fell into the hands of the data scientist, who is by training a software engineer who knows the AI/machine learning technology. This happenstance has created some confusion, frustration, and recreating the wheel. The data scientist is not the business owner or the data resource manager in the organization. That work would be better placed with the “data people,” rather than the software engineer. Table One has some guidelines clarify the specific roles in a AI/ML enabled organization.
What to Know
Supply Chain Scientist
This is the business analyst/supply chain user who is interested in customers, markets, and their environments. This person knows what issues impact supply chains and the sources that can be acquired to analyze those impacts. They intuitively know the semantics about the data, and operate the system.
Within the supply chain team, as well, users may become experts on source data, its value, and impact, for example, how weather affects demand, how trade/import/export regulations affect ocean costs and routes, and so on.
Data Resource Management, the Data Methodology People
They systematize the data about the data (the data dictionary). They know the systemic automation of data ontology and the tools: data lake, databases and the data warehouse technology best suited to serving the varying data needs.
Due to the diversity of data and sources and the applications that might use them, we need to restructure data and enhance its definition based on these new needs. That means we are moving to a world that is beyond the manual creation of relationships in database structures. Data lakes or data fabric (two of the terms you might see) use knowledge graphs, automating the ontology, linking the various types of data stores, processes (programs), and APIs as yet another layer in the information architecture.
AI/Machine Learning Software Engineer aka Data Scientist
The data scientists code and apply machine learning platforms and technologies to search, analyze, and report. They can also automate the ontology, since that may require the use of a host of tools.1
They may also install and support the AI/ML-based applications or work in tandem with the application software expert who supports the supply chain application.
Table One—Roles in the Modern Supply Chain Team
In addition, within the organization other pros have a role to play. The Chief IT Architect, or CTO, is responsible for the technologies and the overall map/architecture of how everything fits together. And we will still have the application-specific software expert who is responsible for the solution and, generally, knowing the business issues as well as the application depth.
It is expeditious, therefore, and kind to think up front about who is going to do what in the new AI/machine learning-powered organization.
The term “what if scenario” is bantered about so much. However, maybe we didn’t stretch ourselves enough when exploring the “what if.” This could be part of the reason why so many companies were caught flat-footed by the dramas we have been subjected to recently.2 And in front of our eyes, more change will definitely happen.
We have to live in our current reality and at the same time, prepare for our future. So, what should we do about this duality? In one philosophical track they advise: do what you must do, but always be open to the possibility that things may be different in the next moment.
How can we create more dynamism in our organization, systems and people, then, so we are prepared for a more fluid world?
Institutionalize discovery and change. Though this may seem like an odd statement—to institutionalize change—but to continue to seek stability as the end-all won’t allow us to see what might be happening right in front of our eyes, or what we need to do next.
Make the processes and IT platform continuous. Break down the artificial barriers between people, data, and how we measure success.
Unleash creativity. Turn to the team for ideas on how to organize for maximum resilience and effectiveness. If we are relying on our people and partners to get it right, we better ask them to design a process where they—and we—can be successful.
Part of the challenge in the human mind and the systems we use is that they are a projection of the past moment, even AI and machine learning systems, to some degree. After all, who is developing and using these? People who rely on their points of view or historical data, which, inherently, are biased. Any bias about what happened before or lack of imagination about what could happen next stymies creativity, and we need to be creative if we are to start asking different questions.