Recently, the Department of Defense has been interested in Artificial Intelligence (AI) and formed several organizations to investigate its potential. Marine Corps logistics leadership has also expressed interest, but to date there has not been a successful demonstration of the technology using Marine Corps data. Those of us who have had to work with Marine Corps data are skeptical that we may run into a “garbage-in, garbage-out” situation if we dive into AI without first addressing the service’s data and information management shortfalls.
Logistics Information Management is an area that requires elevation in our organization and doctrine. We are almost 20 years into the 21st century and nearly 50 years into the information age, yet we have little guidance on how to apply information science to ground logistics management. We might want to consider shoring up our information practices before we take the leap into AI.
The management of logistics information, beginning with data collection, is a critical enabler of modern logistics and supply chain management. It is difficult to communicate its value outside the logistics community and even within the community there is a lot of misunderstanding of the concept, often confusing it with logistics command and control (C2). To frame this discussion, it is helpful to define and discuss a few terms.
First, per the 2019 DOD Dictionary of Military and Associated Terms, C2 is defined as:
“The exercise of authority and direction by a properly designated commander over assigned and attached forces in the accomplishment of the mission.”
The C2 of logistics elements adheres to this definition, but Logistics Information Management is a different science a together. The 2012 NATO logistics handbook* defines it as:
“Logistic Information Management couples available information technology with logistic processes and practices to meet the NATO Commander’s and nations’ logistic information requirements. NATO and nations have numerous users requiring executive, managerial and operational logistic information. To be effective, logistic information systems must facilitate the delivery of the right information to the right people at the right time with the right information security protection. They should cover all logistic functions and interface between these functions and other functional areas as required. NATO logistic systems need to be interoperable with both existing and emerging national and NATO systems. Interfaces with industrial systems should also be considered where practical and cost effective.”
*It should be noted that NATO identifies Logistics Information Management as a function of logistics on par with transportation, supply, maintenance, etc.
The key elements of this definition are:
Couples information requirements and technology
Wide user base from strategic (executive) to tactical (managerial) levels
Timeliness and accuracy of information
Spans all logistics functions
Inter-operable with existing and emerging systems
Interfaces with industry when practical
These elements are central to multiple efforts across the Marine Corps today, but we have not organized the efforts under the central theme of Logistics Information Management, nor have we assigned responsibility to a single owner. This owner would be responsible for setting business rules for the transfer of information regardless of the communications medium. Business rules are standards that determine what we are tracking and what is considered normal performance. This is not the same as governance which falls on C4I.
C4I is responsible for providing the architecture and infrastructure required for Logistics Information Management, but without informed guidance from the logistics community, the systems will be directed at C2 rather than LIM. C4I will not, and should not, be responsible for determining which data and information we choose to communicate in order to inform logistics decisions and management. That responsibility falls squarely on the logistics community and requires a level of understanding of supply chain management that we, quite frankly, have not developed organically.
There is a misconception among many leaders that data exists all over the place and we need to get some sort of IT solution to collect it, interpret it, and tell us answers. Logistics data is structured because it resides in a data base that we control. Which data elements are collected and how it is interpreted is completely up to our logistics leaders. We need to apply information management techniques to the data set and design our processes around them to get timely management insights.
A commonly used visualization for communicating information management is the “information hierarchy.” In this version I have added a technology driven axis and a people/process driven axis. The technology axis starts with data collection which feeds a transactional data base. GCSS-MC is our transactional data base. This data base is archived periodically, usually at least every 24 hours, into a data warehouse. The data warehouses for GCSS-MC are the Master Data Repository (MDR) at LOGCOM, and the Enterprise Ground Equipment Management (EGEM) database at DC I&L. Both are defense contractor operated. Analytic tools can be applied to the historic data contained in the data warehouse(s) to gain insights into operations or performance. AI would fall somewhere even with or above analytic tools on the vertical axis. Unfortunately, results are limited due to several factors that could be mitigated by investing in the horizontal axis.
The horizontal axis is all about management, starting with the transactional user. The enterprise must enable the entry level user to collect accurate and complete data. Each level of management has differing information requirements, informed by elements lower on the curve determined by well-defined business rules. As you move up the curve, each level has a wider perspective of the data and information. The resulting diagonal curve is the information hierarchy that moves us from raw data to understanding. This model is incomplete without a feedback mechanism to improve data collection and information management processes.
The Marine Corps has invested heavily in this technology axis but has not cracked the code on making work to the level expected by leadership. There must be a corresponding investment in people and enterprise wide processes to move up the information hierarchy. The Marine Corps will never make it into the understanding block with its current approach. Investments in AI at this point are premature. AI for structured data is rules-based. It takes people and processes to establish the rules before the AI can be employed to make performance-based decisions.
There should be an incremental approach to implementation of a solution that marches us toward the ability to fully capitalize on our technology investments. In general efforts could be organized under three lines of effort:
1) Enable People:
Improve professional education in relevant, information age, logistics fields
Provide tools, sensors, and apps to enable the transactional user to complete their task in the most expeditious manner possible while seamlessly collecting required data to populate KPI’s.
2) Develop and Improve Processes:
Manage the information and establish business practices that incorporate a feedback loop for continuous improvement
Develop KPI’s adapted from industry to incentivize and achieve desired results.
Employ KPIs at the unit level for management to monitor their own performance. As a side benefit of employing these tools at a low level in the organization, is that the units will investigate and correct outliers, thereby cleaning the data before it progresses up the information hierarchy.
Develop apps oriented on the transactional user. Separate the users from direct interaction with the transactional database.
Develop dashboards to display KPI’s for different user types and different levels of the organization.
Integrate information systems across the Joint Logistics Enterprise (JLEnt) and enable near real-time analysis.
The Marine Corps logistics IT investment strategy is missing the people and process elements that would make it successful. There is no single owner of the problem set for the logistics community and as such many well-intentioned efforts are disaggregated, sporadic, and doomed to failure. Only through establishing a central owning agency within the logistics community and adopting a measured approach, informed by knowledgeable people, and continuously improved, will we achieve a level of minimum performance that is requisite for any application of rules-based AI.