GET ON THE AI TRAIN: Appropriate training environments and tools are essential to successfully integrating AI. The GCSS-Army virtual training environment could be a tailorable platform used to train Soldiers to apply AI tools and existing data. (U.S. National Guard photo by Master Sgt. Becky Vanshur, 124th Fighter Wing)
How Army sustainment can lead the integration of artificial intelligence for the Army and DOD.
by Maj. Nathan Platz and Maj. Andy Horn
Imagine a world where commanders make decisions with the aid of predictive models informed by real-time data. Where staffs no longer aggregate outdated data and build PowerPoint slides but focus instead on assessing information and optimizing processes to reduce waste. This is the potential and more for artificial intelligence within the Army and DOD. We believe that the Army sustainment enterprise is postured to lead the integration of artificial intelligence (AI) within the Army and DOD because of a large volume of data available through Global Combat Support System-Army (GCSS-Army). The Army and DOD have recently prioritized implementing AI into the military operations; however, a 2019 RAND Corp. study, “The Department of Defense Posture for Artificial Intelligence,” highlights existing challenges in “organization, advancement, adaption, data and talent.”
That study outlines five areas for change within DOD to fully harness AI as a tool for providing predictive analysis for commanders to improve decision-making. AI is a general term for a program that learns a process from existing data and makes a prediction about a future state. The wealth of data in GCSS-Army is the perfect environment for Army sustainment to lead DOD in adopting AI tools for application into military operations.
STEP I: DATA
Become diligent about data collection and begin training AI tools against currently available data.
As summarized in the book “Prediction Machines,” by Ajay Agrawal, Joshua Gans and Avi Goldfarb, AI is “prediction technology” where prediction is the process of filling in missing information—prediction takes information you have, often called “data,” and uses it to generate information you do not have.
The first step to integrating AI in the Army is to apply AI toward current data sets. Data is the key—the best algorithm will fail if the data is inconsistent or sparse. RAND found that the advances in AI “are predicated on the availability of large labeled data sets and significant computing power to train the algorithms.” The Army sustainment enterprise is ahead in DOD’s and the Army’s integration of AI because of data available in GCSS-Army, which, according to Northrop Grumman, is the largest enterprise resource planning system ever successfully attempted within DOD. As a result, the Army has large, labeled data sets in GCSS-Army for maintenance, materiel management and supply.
The Army needs to develop algorithms to train against this data and apply AI to provide predictions on future maintenance and supply status from the tactical to the strategic level. The Army needs to begin training the algorithms to provide predictive analysis to commanders now. The benefit of this is to enable commanders to assess the usefulness of these predictions and forecasts in making decisions, further improving the application of AI. This offers a source of data for training, verification and testing to hone algorithms and to produce models of predicted future outcomes. The Army could use the National Training Center at Fort Irwin, California, and the Joint Readiness Training Center at Fort Polk, Louisiana, as initial hubs for training and validating algorithms on the existing data. Personnel would see real-time status of logistics assets and commanders could make decisions informed by predictive analysis from that data. As data is collected and decisions made from that data, the algorithms would become more accurate and the predictions more precise. The number of Army units that rotate through these centers, the controlled environment and sophisticated data collection tools already in existence make these prime environments for testing early AI integration on existing data.
STEP 2: COLLECTION
Improve and expand data collection efforts to integrate the tracking of ammunition, fuel and transportation.
While the data in GCSS-Army establishes a great base for applying AI tools, the Army requires improvements and expansion to data fed into that database. The Army is leaning forward in this endeavor through expanding GCSS-Army. In a 2017 article in Army Sustainment magazine, “GCSS-Army: Providing big data for readiness,” the commander of the Combined Arms Support Command, Maj. Gen. Rodney D. Fogg, discussed a future addition to GCSS-Army that would include the ability to track ammunition, fuel and transportation. Providing the visibility offered for maintenance and supply in GCSS-Army to other areas of logistics is essential. This demonstrates that the Army sustainment enterprise has a foundation on which to build AI platforms.
The Army must integrate this capability and the data collection efforts now. While it is largely agreed upon that the implementation of GCSS-Army was successful, the transformation took 10 years: six years of operational assessment at Fort Irwin from 2007-2013, and four years for phase I and phase II integration from 2013-2017. A similar timeline is likely to integrate data collection for transportation, fuel and ammunition. Establishing a consistent data set and the data collection process for these areas of logistics is crucial. RAND found that a hindrance to AI integration is that data is not collected and stored at every opportunity. This is currently true for transportation, ammunition, fuel, and water—this is the hurdle the Army must clear to integrate tracking of commodities in GCSS-Army and to apply AI to inform commanders across all facets of logistics. Once the Army is able to consistently collect data from all areas of logistics and integrate that data into GCSS-Army, the Army can apply AI tools against the complete logistics dataset and offer timely predictive analysis to commanders at echelon.
STEP 3: STANDARIZED FORMAT
Standardize the Logistics Common Operating format at echelon and have it fed by cloud based data.
Logistics commanders from the tactical to the enterprise track metrics on an logistics common operating picture to inform decisions and help see the enemy and themselves. The formats for these logistics common operating pictures are not in fact, common, but are all unique to each unit despite displaying the same information. The lack of standardized format forces staffs to spend hours formatting PowerPoint and excel documents to track latent data. The Army has a great tool in the Army Readiness-Common Operating Picture with potential to advance the integration of informing decisions utilizing AI. Army Readiness-Common Operating Picture displays information from GCSS-Army logically to allow leaders to “see themselves.” While Army Readiness-Common Operating Picture is a great visual tool, it only displays historical data for maintenance and supply without providing useful predictive analysis and it is not useful at the tactical level.
As described by Fogg in his 2019 article “Building the Army Readiness-Common Operating Picture”, Army Readiness-Common Operating Picture needs to provide commanders with analytics. We think those analytics must be fully informed by AI and useful at echelon. To do so, we need to incorporate the expansion data from additional logistics commodities in GCSS-Army and apply AI against the data for predictive analysis in Army Readiness-Common Operating Picture to inform decisions. This will enable commanders timely predictions and forecasts rooted in data and staffs will have the ability to focus on the future state rather than working to confirm the current state.
STEP 4: EXPAND TRAINING
Expand training for enterprise resource platforms and data analysis at all levels of professional military education.
In a recent meeting on the future application of data within United States Special Operations Command, Chief Data Officer Thomas Kenny briefed that the Defense Logistics Agency saved 130,000 man-hours last year by automating all displays. This removed the requirement for briefing on PowerPoint and freed staff up to conduct analysis.
Key to successfully integrating AI is establishing a training environment and training tools similar to Anaconda (an open-source programing environment) and Jupyter Notebook (open-source software for interactive computing) platforms for Python programming language. The Army sustainment enterprise can expand on the existing GCSS-Army virtual training environment to enable Soldiers to apply AI development tools against data in GCSS-Army and share code.
This would allow a scalable and tailorable training platform to apply AI tools to existing data. Existing GCSS-Army business intelligence training platforms and tools should be elevated beyond their current optional state of learning to include AI applications once integrated in the GCSS-Army and Army Readiness-Common Operating Picture (AR-COP) environment. Along with the training environment, the Army must establish training standards for data as it improves data collection efforts.
In addition to training, the Army must leverage civilian expertise through direct hires and expanding existing partnerships with leaders in the AI space—Microsoft, Amazon, etc. The Army should not overlook the warrant officer ranks in its development of AI expertise. Warrant officers already lead efforts in understanding GCSS-Army application and should do the same leading the integration of AI.
STEP 5: THE ACQUISITION PROCESS
Leverage the acquisition process to enhance data collection capability and improve AI integration.
The final area of consideration is the American industrial base and our acquisition process. Leveraging the acquisition process is an advantage available to the entire Army and DOD. The Army must leverage its power in procurement for the cultivation of data and future integration of AI. As we previously described, we lack consistent data collection devices for fuel, water, transportation, and ammunition—yet these devices exist in the commercial sector. On the commercial side, we can use an app on our phone to monitor the fuel level in our personal vehicle or track a package as it moves form Amazon to our house. The Army sustainment enterprise needs to use the acquisition process and relationships with suppliers to demand similar data collection capabilities in all products, and most importantly, require access to data on all equipment from defense contractors.
The Army needs to apply the government contract process to prioritize manufacturers and developers who embrace AI and incentivize those without AI to adopt data collection capabilities that support the Army and DOD integration of AI. The Army must focus its attention as much on software as it does on hardware in order to keep pace with advancing technology.
Through initiatives underway to expand data fed into GCSS-Army and the maturation of the Army Readiness-Common Operating Picture, the Army sustainment enterprise has established the initial framework for the development and integration of AI. This framework is in line with DOD goals to provide “AI-enabled information, tools and systems to empower, not replace, those who serve.” Applying AI framework to existing data to provide predictive analysis for commanders through AR-COP, improving AI training nested with GCSS-Army and leveraging the industrial base to assist with data collection and AI integration are the next steps. These steps are steep and the path is littered with obstacles, but the Army sustainment enterprise is primed to lead the change in AI for the Army and DOD. These steps are only the first of many required for us to realize that imagined world.
“Advances in AI have the potential to change the character of warfare for generations to come. Whichever nation harnesses AI first will have a decisive advantage on the battlefield for many, many years. We have to get there first.”
—Secretary of Defense Mark T. Esper.
For more information, contact the authors at email@example.com or firstname.lastname@example.org.
MAJ. NATHAN PLATZ is a logistics officer currently serving as the deputy director of logistics (J-4) for Special Operations Command North, Peterson Air Force Base, Colorado. He has served in Airborne and Stryker brigade combat teams, sustainment brigades and in special operations with experience in Iraq, Afghanistan, Europe and the Pacific. He holds an MBA from the College of William and Mary and a B.S. in computer science from Missouri State University.
MAJ. ANDREW HORN is a logistics officer currently serving as the executive officer to the commanding general of 8th Theater Support Command, Fort Shafter, Hawaii. He has served in infantry brigade combat teams, sustainment brigades and in special operations with experience in Iraq, Afghanistan, Europe and the Pacific. He holds an M.S. in supply chain management from the University of Kansas and a BBA from Pacific Lutheran University.