In 1950, Brigadier General S.L. Marshall, U. S. Army Reserve, wrote “The Soldier’s Load and the Mobility of a Nation.” Though it was written well before the term machine learning (ML) existed, Marshall’s observations of men at war and the role of supporting logistics is highly readable and relevant today, even in this time of rapidly growing computerized autonomy in defense organizations as well as other industries. The purpose of this blog is to connect the dots between Marshalls’ observations with today’s technology—or at least generate some thought on this subject.
Marshall explains that fatigue, fear, and physical load reduces the ability to act. Defense and other industries are impacted by these factors and require a distinct perspective on the potential for ML. Marshall wrote, “It seems to be taken for granted that the introduction of the machine into warfare is tending to produce automatic solutions of the prevailing problem of how to get more fire out of fewer men.”
As Marshall’s comments are directed at combat power, he is probably referring to weapons systems. But what if we move his comments into the 21st century and shift the discussion to include combat support/combat service support as well? Then he might say something like this: “It seems to be taken for granted that the introduction of business systems into defense organizations can produce automatic solutions of the prevailing problem of how to get more combat support out of fewer people.”
In summary, combat support and combat service support troops suffer reduced effectiveness in combat because of fatigue and uncertainty (and sometimes fear). Routine tasks can become confusing or fail to be accomplished. And some of the tasks currently performed by humans can be supported by ML or done better by autonomous systems.
As experiences in Iraq and Afghanistan have recently reminded us, a combat environment may not have a clear distinction between the front and rear areas. A mobile contact/mobile repair team supporting an armored or aviation unit is subject to the same physical and psychological factors as traditional combat personnel and combat forces initiate or monitor logistics actions like ordering ammunition, food/water resupply, or medical treatment—so traditional combat troops are now involved in logistics processes to some degree.
The question remains: How can business systems facilitate the conduct of these forces in the performance of their mission and improve their readiness? Said another way, how can machine learning reduce the soldier’s load?
First, let’s look at what ML is. One of my favorite definitions is from TechEmergence:
“Machine learning is the science of getting computers to learn and act like humans do, and improve their learning over time in autonomous fashion, by feeding them data and information in the form of observations and real-world interactions.”
In general, there are several key drivers behind the march of ML:
- The proliferation of Big Data: Datasets are magnitudes larger today than they used to be and they come from sources not imagined in Marshall’s time—every weapon system, generator, and mobile device generates data.
- Better algorithms exist today and are improving every day. These algorithms can be bought, downloaded, used as shareware or freeware, or custom-written for a specific purpose. These algorithms are a key factor in what enables a computer to “learn.”
- Massive computing power and the proliferation of computing power in more devices and accessible via the cloud.
The Department of Defense is aware of both the impact and the opportunity. In November 2014, U.S. Secretary of Defense Chuck Hagel announced the U.S. DoD’s intentions to lead the coming autonomy-driven military-technical revolution. In FY2017, the DoD spent over $2 billion on these technologies.
ML is allowing humans to be more productive and to focus on higher-value tasks. Commercial companies and militaries see ML as critical for competitive advantage. Because much of the research and development for ML is being done in the commercial sector, products are available to all, and there will be ease of access and a high degree of competition for resulting benefits from these tools.
What are the situations where machine learning and business systems may help reduce the soldier’s load?
- A soldier or sailor uses a digital assistant for reporting and data entry via voice instead of writing or typing.
- An airman takes photos to identify broken parts and orders them to repair an aircraft on the ground. (Computers’ photo recognition skills are now better than a human’s.)
- A Marine gets recommendations on what to order while conducting unit replenishment of supplies: “I think you need batteries for your laser rangefinder. Please confirm and I’ll order for you.”
- A vehicle maintainer gets a suggested task list for work orders based on identified broken parts: “While you’ve got that alternator pulled out, look at starter behind it too. It’s approaching end of life based on information its sensors are providing.”
High-value areas of ML for the defense community include connected fleet management, dynamic inventory optimization, dynamic maintenance management, predictive maintenance & services, and warranty analytics.
In summary, some estimates suggest that by 2025, 60 percent of human tasks will be automated. Today’s business systems already can leverage ML, and these systems are highly relevant in environments where fatigue or other factors reduce human effectiveness; or where limited human resources suggest using autonomous systems for some tasks and humans for others.
So, what to do next? Gartner is telling customers to do the following in the near term:
- Develop your employees and get them used to the idea of ML in some select areas.
- Develop this initial understanding into deeper understanding by piloting initiatives in areas where there is the opportunity to mine data quickly
- Exploit software features being introduced by your ERP vendor—plow lessons learned into future planning.
These are good recommendations, and I recommend choosing from the list of high-value opportunities listed above.
For more on emerging technology in the defense industry, see Disconnected Operations In Defense: Identifying The Right Business Processes.
Written by two infantrymen: Mike Lennon and Matthias Ledwon