Over the past year, we have heard much talk about the supply chain. This blog will discuss the supply chain and how AI/ML can help resolve the current challenges and prevent future breakdowns.
First, we should begin with the definition of supply chain management. Supply Chain Management is the sequence of business processes that produce and distribute a commodity for a competitive advantage. For example, building a microchip requires a supply chain comprised of many companies. First, a company finds and harvests the raw material (silicon, metals, etc.). The silicon then needs to be transported to a manufacturing company that can make the semiconductor that a separate company may have previously designed and is selling the patent to the manufacturer. Finally, once the microchip has been produced, it needs to be packaged and stored before being transported and sold to the customer.
Supply Chain Management should not be confused with logistics, which manages the flow and storage of goods, services, and information. Logistics is an activity within Supply Chain Management, and thus, the two terms should not be used interchangeably.
To enhance a supply chain and maintain a competitive advantage, a company may lock down one or more producers. Examples would be mining or manufacturing raw materials, purchasing patents, purchasing or creating a delivery company or methodology. Although one may try to control the logistics, much of that is outside the producer’s control. These products will often rely on commodity logistics services such as the USPS, FedEx, UPS, or large freight shipping companies like P. Moller–Maersk or Mediterranean Shipping Company to get from the manufacturing facility to the consumer.
I am often asked how we arrived in a global supply chain crisis. In short, several issues have significantly contributed to the current supply chain quandary. The primary reason explanation is the global Covid-19 pandemic. Many companies shut down their services for several weeks during the pandemic, hoping that this virus could be contained and eradicated via a quarantine. The service disruption caused manufacturing in many industries to almost come to a complete stop. For example, 70% of the world’s microprocessors get created by a single company in Taiwan. This company stopped manufacturing at times during the pandemic to slow the spread of the coronavirus. While the manufacturing industry stalled, the demand for products, such as computers, monitors, and other home office essentials, skyrocketed as many sectors, including government agencies, tried to transition to a work-from-home solution.
Supersized cargo ships loaded with consumer electronics and other goods were now trying to meet the demand and unload their freight at ports, but there were issues at the docks. The ports quickly became overcrowded due to everything from shortages of dockworkers and equipment to a lack of truck drivers to haul away the cargo containers. The result was a long queue of ships waiting to get into the port and unload their freight. Customers feel this impact via empty store shelves, delayed deliveries, and increased cost of goods and services.
The version I described above is simplified to what happened in 2020 and 2021. I have ignored the local and global political and social issues that also contributed to the supply chain issues. With this high-level overview as a background, how can AI/ML help resolve the current crisis and prevent a future calamity from reoccurring?
At BigBear.ai, we believe that AI/ML can help every aspect of the Supply Chain Management process. For example, in BigBear.ai’s commercial sector, we use our VANE product to help maritime shipping companies optimize product flow and ship routing to avoid the port congestion issues we are experiencing today. These companies are also using this AI/ML information to buy and sell futures contracts on these vessels weeks and months before they are scheduled to sail. BigBear.ai’s VANE software also optimizes the maritime shipping process by tracking manifests, weather, port facilities, and dozens of other variables to get the ship to the best port at the right time. How good is sending goods to a port that cannot unload the products for days or weeks? A delay lasting more than a few hours can cost the shippers hundreds of thousands of dollars per day.
VANE’s AI/ML capabilities can be used to prescribe alternative solutions, like unloading at a different port that is available to receive the ship and cargo. For example, if the port of Savanah is congested or there are not enough truck drivers available to haul away the cargo, can the ship dock in New Orleans, Houston, or Galveston instead to minimize the delay and optimize the profit?
In the federal sector, we are starting to see an uptick in agencies accepting and implementing AI/ML solutions to address the information flow challenges we see plaguing the information supply chain in the Department of Defense (DoD) and Intelligence Community (IC). Our unique ability to clean and derive data using tensor completion technology enhances the accuracy of our AI/ML solutions. Not only does VANE make better predictions, but it also provides our commercial and federal customers with the possible repercussion of each course of action. This prescriptive information allows the decision-maker to make the best possible decision based on what they know at that time.
AI/ML is not a panacea for decision-making. However, it is another tool in your arsenal that can help make the best decision possible while helping one understand the ramifications of the possible courses of action. For example, a few years ago, I was a part of a team that improved event predictions from 20% the day before the event to 80% seven days in advance of the event using our VANE technology. However, although 80% is a high number, there is still a 20% possibility of a false-positive, or an incorrect prediction. Therefore, one should never eliminate the human from this process when lives or large sums of money are at stake, as an erroneous result could have devastating consequences.
In conclusion, the data remains the linchpin for any AI/ML solution. Although you will need high-quality data scientists to build insightful models, you must have a robust set of relatively clean and complete datasets to find unique insights and make accurate predictions about your area of responsibility. Without the data, you cannot make a meaningful prediction.
About the Author
Jim McHugh is the Vice President of National Intelligence Service – Emerging Markets Portfolio. McHugh is responsible for the development and delivery of technical solutions for Analytics and Data Management. His responsibilities include defining and implementing standards consistent with current and future technology trends, assuring adherence to standards and good development techniques, mentoring, and project management.
