Chris Oleksy, Founder and CEO, Oleksy Enterprises and Co-Founder/CEO, Next Life Medical; CEO, Emergent Respiratory04.01.22
It’s time to dispel the myth.
For years now, there’s been a folk tale floating around the medtech industry that professes artificial intelligence (AI) will make the supply chain profession obsolete. Not true. In fact, quite the opposite is likely to occur: The world will need more sophisticated talent, not less, to leverage AI’s cornucopia of benefits. I’ve had significant experience with AI dating back 35 years (indeed, the technology has been around for that long) and I’ve seen a similar myth dispelled in another industry.
Being raised in a bank family in the 1970s, I can distinctly remember when ATMs were being placed in banks and every public location known to man. Many people at the time thought—and some secretly hoped—that ATMs would eliminate tellers. It was very difficult to find qualified tellers who could balance their cash drawers to the penny on a daily basis. Since most transactions were conducted with cash, tellers had to accurately balance their drawers in order for the bank to collectively balance its assets. Thus, automating cash transactions was logical. But the banking industry knew ATMs would create easy access to cash, as the term “charge it” was rarely heard at the time. Cash was king, so, it needed to be available practically everywhere. Conventional wisdom dictated that fast, easy cash accessibility would translate into increased consumer spending.
The “attempted transition” from teller to ATM spawned sophisticated electronic technology that safely linked ATMs to banks worldwide. During this process, the financial industry realized that credit could be made more readily available through machines as well. Consequently, the banking industry’s evolution to ATMs triggered a need for more sophisticated tellers performing different tasks. Tellers were needed to help convince and educate consumers about the merits of credit use—namely, its ease, accessibility, security, and superiority to cash, not to mention free airline flights. Ultimately, ATMs helped move the financial industry to the next level, but rather than eliminate tellers the evolution redefined their roles by transforming them into personal bankers. The same evolution is occurring within the supply chain via AI technology.
Today’s supply chain domain is very similar to the 1970s banking industry. Like tellers, it’s hard to find talented supply chain professionals who can maximize the efficiency of an organization during good times and, more importantly, tough times. Thus, in the absence of available talent, it would seem logical to automate or possibly eliminate many supply chain roles. It is a well-known fact that supply chain activities comprise 35 to 50 percent of business costs, so these roles are critical. Although many supply chain tasks are trivial and mathematical in nature—making them quite conducive to AI technology—they are not easily eliminated.
I started my career in the mid-1980s. Having studied quantitative business analysis and production operations management in college (equivalent to a supply chain degree today), I joined Dow Corning, which was ahead of its time and eager to take its supply chains to the next level. At the time, supply chains were called “quality circles”—at least until the Supply Chain Operations Reference Model (SCOR) was born. Regardless of the name, the goal was the same—connect all constituents to maximize an organization’s offering in an economical, qualitative manner. Due to the amount of tactical/trivial activities within supply chain roles, I suggested that Dow Corning explore a new, evolving area in operations research called Expert/Knowledge-Based Systems, which was later known as AI systems. I knew this evolving technology could help automate many tactical activities. Coincidentally, at that same time, MIT was exploring a “new area” called artificial intelligence (AI) and looking for large manufacturing partners to participate in research. A forward-thinking management team at Dow Corning gave me the green light to work with MIT.
During this fascinating time, I learned something that forever changed my view on technology use—it is only as good as the person using it. With AI technology, however, there is a subtle but extremely important difference. AI technology is not something used, but rather something built. Expert knowledge, in the form of rules, must be built into the technology so it can function. Unlike typical technology that is simply used, AI functions based on instructions from the user. The more expert the user, the more expert the results. Unfortunately, the opposite is true—a weak user will produce weak results.
Consider the concept of email, for example. Email is a technology that is used; it does not think or make decisions on anyone’s behalf, or help solve problems. It is simply pre-developed software used for communication purposes. Conversely, AI technology has built-in rules that enable it to either make decisions or offer a suggested path.
To purchase raw materials over a future period of time, sourcing professionals employ a concept called Materials Requirement Planning (MRP). MRP is a mathematical process that takes into consideration many variables such as inventory, estimated consumption vs. demand/forecast, and product intake from suppliers to provide a Projected Available Balance (PAB) over a certain period of time. The software provides “mathematical advice” such as when to buy a specific quantity of material. Most MRP systems are not set up to automatically buy raw materials unless they have been instructed to do so. That instruction is a rule based on knowledge, and that knowledge comes from an individual or agreed upon strategy—all of which are created by humans. In this example, that individual would be a supply chain sourcing expert. Simply put, AI is nothing more than the use of rules created by knowledgeable experts and then executed by software. Thus, if MRP software is only as good as the person using it, MRP-AI software is only as good as the person instructing it.
The newest hot topic in AI is machine learning, where instead of a human creating the software rules, the software itself is instructed to learn from success and failures and automatically update its rules. For example, a built-in financial objective could teach software to reduce inventory levels if raw material supplies remain steady over a certain period of time, or automatically increase inventory levels when supplies are running low. It would learn from success/failure and adjust accordingly.
One issue that severely hampers the ability to effectively leverage AI is black swan events—unpredictable, unexpected incidents with potentially severe consequences beyond what is normally expected of a situation. Since rules are controlled by users or the machines themselves based on events that are normally expected, rules need to have a degree of predictability to them. Recent black swan events such as COVID-19 and Russia’s invasion of Ukraine are far from predictable and seem to be occurring at a much more frequent pace lately.
Another area of major concern with AI—as with any digital system—is cyberattacks. These systems can easily become vulnerable to hacking because they are exposed to the outside world. AI rules could effortlessly be rewritten to act upon something that was not originally intended, producing an inadvertent outcome. A cyberattack, for example, could very well turn a routine 10-pallet raw material order into a 100-pallet headache.
AI technology will certainly have its place throughout various supply chain activities across the Plan-Source-Make-Deliver (SCOR) continuum. But its success depends largely on the level of qualified supply chain talent available to harness it. AI will likely reposition efforts consumed on supply chain tactical activities towards more sophisticated initiatives. Like tellers becoming personal bankers, it will demand a much higher level of sophistication among supply chain professionals, which today are in short supply. In addition to learning the science of performing activities such as MRP, future supply chain professionals will need more training on the art of the value chain known as rules. Part of that art will require knowing the most appropriate times and situations in which to leverage AI. The other part is realizing the system is only as smart as its (human) teacher.
Chris Oleksy is founder and CEO of Oleksy Enterprises, co-founder/CEO of Next Life Medical, and CEO of Emergent Respiratory. He can be reached at chris@oleksyenterprises.com or chris@nextlifemedical.com.
For years now, there’s been a folk tale floating around the medtech industry that professes artificial intelligence (AI) will make the supply chain profession obsolete. Not true. In fact, quite the opposite is likely to occur: The world will need more sophisticated talent, not less, to leverage AI’s cornucopia of benefits. I’ve had significant experience with AI dating back 35 years (indeed, the technology has been around for that long) and I’ve seen a similar myth dispelled in another industry.
Being raised in a bank family in the 1970s, I can distinctly remember when ATMs were being placed in banks and every public location known to man. Many people at the time thought—and some secretly hoped—that ATMs would eliminate tellers. It was very difficult to find qualified tellers who could balance their cash drawers to the penny on a daily basis. Since most transactions were conducted with cash, tellers had to accurately balance their drawers in order for the bank to collectively balance its assets. Thus, automating cash transactions was logical. But the banking industry knew ATMs would create easy access to cash, as the term “charge it” was rarely heard at the time. Cash was king, so, it needed to be available practically everywhere. Conventional wisdom dictated that fast, easy cash accessibility would translate into increased consumer spending.
The “attempted transition” from teller to ATM spawned sophisticated electronic technology that safely linked ATMs to banks worldwide. During this process, the financial industry realized that credit could be made more readily available through machines as well. Consequently, the banking industry’s evolution to ATMs triggered a need for more sophisticated tellers performing different tasks. Tellers were needed to help convince and educate consumers about the merits of credit use—namely, its ease, accessibility, security, and superiority to cash, not to mention free airline flights. Ultimately, ATMs helped move the financial industry to the next level, but rather than eliminate tellers the evolution redefined their roles by transforming them into personal bankers. The same evolution is occurring within the supply chain via AI technology.
Today’s supply chain domain is very similar to the 1970s banking industry. Like tellers, it’s hard to find talented supply chain professionals who can maximize the efficiency of an organization during good times and, more importantly, tough times. Thus, in the absence of available talent, it would seem logical to automate or possibly eliminate many supply chain roles. It is a well-known fact that supply chain activities comprise 35 to 50 percent of business costs, so these roles are critical. Although many supply chain tasks are trivial and mathematical in nature—making them quite conducive to AI technology—they are not easily eliminated.
I started my career in the mid-1980s. Having studied quantitative business analysis and production operations management in college (equivalent to a supply chain degree today), I joined Dow Corning, which was ahead of its time and eager to take its supply chains to the next level. At the time, supply chains were called “quality circles”—at least until the Supply Chain Operations Reference Model (SCOR) was born. Regardless of the name, the goal was the same—connect all constituents to maximize an organization’s offering in an economical, qualitative manner. Due to the amount of tactical/trivial activities within supply chain roles, I suggested that Dow Corning explore a new, evolving area in operations research called Expert/Knowledge-Based Systems, which was later known as AI systems. I knew this evolving technology could help automate many tactical activities. Coincidentally, at that same time, MIT was exploring a “new area” called artificial intelligence (AI) and looking for large manufacturing partners to participate in research. A forward-thinking management team at Dow Corning gave me the green light to work with MIT.
During this fascinating time, I learned something that forever changed my view on technology use—it is only as good as the person using it. With AI technology, however, there is a subtle but extremely important difference. AI technology is not something used, but rather something built. Expert knowledge, in the form of rules, must be built into the technology so it can function. Unlike typical technology that is simply used, AI functions based on instructions from the user. The more expert the user, the more expert the results. Unfortunately, the opposite is true—a weak user will produce weak results.
Consider the concept of email, for example. Email is a technology that is used; it does not think or make decisions on anyone’s behalf, or help solve problems. It is simply pre-developed software used for communication purposes. Conversely, AI technology has built-in rules that enable it to either make decisions or offer a suggested path.
To purchase raw materials over a future period of time, sourcing professionals employ a concept called Materials Requirement Planning (MRP). MRP is a mathematical process that takes into consideration many variables such as inventory, estimated consumption vs. demand/forecast, and product intake from suppliers to provide a Projected Available Balance (PAB) over a certain period of time. The software provides “mathematical advice” such as when to buy a specific quantity of material. Most MRP systems are not set up to automatically buy raw materials unless they have been instructed to do so. That instruction is a rule based on knowledge, and that knowledge comes from an individual or agreed upon strategy—all of which are created by humans. In this example, that individual would be a supply chain sourcing expert. Simply put, AI is nothing more than the use of rules created by knowledgeable experts and then executed by software. Thus, if MRP software is only as good as the person using it, MRP-AI software is only as good as the person instructing it.
The newest hot topic in AI is machine learning, where instead of a human creating the software rules, the software itself is instructed to learn from success and failures and automatically update its rules. For example, a built-in financial objective could teach software to reduce inventory levels if raw material supplies remain steady over a certain period of time, or automatically increase inventory levels when supplies are running low. It would learn from success/failure and adjust accordingly.
One issue that severely hampers the ability to effectively leverage AI is black swan events—unpredictable, unexpected incidents with potentially severe consequences beyond what is normally expected of a situation. Since rules are controlled by users or the machines themselves based on events that are normally expected, rules need to have a degree of predictability to them. Recent black swan events such as COVID-19 and Russia’s invasion of Ukraine are far from predictable and seem to be occurring at a much more frequent pace lately.
Another area of major concern with AI—as with any digital system—is cyberattacks. These systems can easily become vulnerable to hacking because they are exposed to the outside world. AI rules could effortlessly be rewritten to act upon something that was not originally intended, producing an inadvertent outcome. A cyberattack, for example, could very well turn a routine 10-pallet raw material order into a 100-pallet headache.
AI technology will certainly have its place throughout various supply chain activities across the Plan-Source-Make-Deliver (SCOR) continuum. But its success depends largely on the level of qualified supply chain talent available to harness it. AI will likely reposition efforts consumed on supply chain tactical activities towards more sophisticated initiatives. Like tellers becoming personal bankers, it will demand a much higher level of sophistication among supply chain professionals, which today are in short supply. In addition to learning the science of performing activities such as MRP, future supply chain professionals will need more training on the art of the value chain known as rules. Part of that art will require knowing the most appropriate times and situations in which to leverage AI. The other part is realizing the system is only as smart as its (human) teacher.
Chris Oleksy is founder and CEO of Oleksy Enterprises, co-founder/CEO of Next Life Medical, and CEO of Emergent Respiratory. He can be reached at chris@oleksyenterprises.com or chris@nextlifemedical.com.