21st Century Herodotus: Developing Future Artificial Intelligence Leaders Today
From cyber warfare to drone swarms and ubiquitous sensors, artificial intelligence and its subset of machine learning are predicted to become key components in many of the warfighting technologies of the future often touted as the U.S. military’s Third Offset Strategy. However, near-peer competitors are also pursuing military artificial intelligence and machine learning technologies to gain a comparative warfighting advantage. Engagements with such competitors would likely pit U.S. military artificial intelligence against adversary artificial intelligence technologies, resulting in a contest between machines for an advantage. Modern and future competition therefore, introduces an element of dependence on outcomes of inscrutable algorithms reliant on cloud-level computing for potentially determining the fate of the force. Not surprisingly, these advanced military technologies are shaping up to provide nothing short of a Homeric epic for future militaries.
Greek mythologies, while not perfect analogies, provide ample cautions for military leadership faced with the prospect of future algorithmic warfare. Advanced military technologies named after Greek mythological characters—Harpy, Gorgon, Athena, Aegis, Talos, etc.—suggest an analogical construct reminiscent of ancient heroes who relied on the favor of the gods to tip the enigmatic scales in their favor. Similarly, the results of employing a military’s autonomous machinery against equivalent adversarial technologies and strategies will likely be unknowable or difficult to judge until an engagement takes place. For this reason, enhancing professional military education, democratizing experimentation, and improving collaboration will go far towards improving individual warfighter abilities to wield these technologies effectively, which may provide the only measurable advantage in future warfare.
Borrowing from the Greeks
There is no denying that modern artificial intelligence and machine learning platforms pay homage to Greek mythology. Examples include: Minotaur, a combined Navy and Marine Corps system for automated data analysis; Gorgon Stare, an Air Force program that uses a series of sophisticated cameras for intelligence, surveillance, and reconnaissance that will eventually use the labeling algorithms being refined by the Department of Defense’s pathfinder artificial intelligence program, Project Maven; or Athena, a wargaming environment for capturing game data to inform the development of future artificial intelligence applications. References to artificial intelligence being equivalent to opening Pandora’s Box or having an Achilles Heel and human-machine teaming allusions to Centaurs provide a wide cast of mythological analogs for investigating future warfare. Yet, more importantly, using artificial intelligence technologies could result in algorithms fusing sensor data and real-time wargame results to produce courses of action commanders may come to rely upon for making critical decisions. As a result, it is easy to see how underlying artificial intelligence and machine learning technologies could become the modern Oracle of Delphi for future commanders.
Building Apollo’s Temple
Today’s most sophisticated artificial intelligence and machine learning technologies are built on cloud infrastructure. Cloud-based technologies provide a way for organizations to quickly provision scalable and cost-effective computing resources. Because artificial intelligence and machine learning applications can require significant amounts of storage and computing power, it is no wonder migration to a secure cloud will likely become a prerequisite for the most advanced military artificial intelligence and machine learning applications. For instance, the most sophisticated artificial intelligence and machine learning applications Alphabet, Amazon, and Microsoft offer are built on their cloud infrastructures. Most importantly, these technologies offer a way for any organization to quickly provision storage, compute power, and use of pre-trained cognitive services or natural language processing for rapid employment of applications. The Department of Defense’s move towards contracting cloud services presages a future in which many applications and data storage are located in the commercial cloud. More importantly, the establishment of the Joint Enterprise Defense Infrastructure and the Joint Artificial Intelligence Center creates a symbiotic relationship where big data analytics will power future algorithmic warfare. However, the U.S. military is not the only country pursuing the advantages of cloud computing to support artificial intelligence and machine learning.
Both Russia and China are investing significantly in cloud infrastructure. The Russian news outlet Izvestia reported its country’s armed forces would receive a “closed ‘cloud’” for proprietary and confidential information. Similarly, China’s quest for Informatized Warfare leverages big data and cloud computing to pursue artificial intelligence. China is also pursuing an artificial intelligence policy of military-civil fusion. It will therefore be unremarkable if China’s military leverages one of their leading civilian firm’s Apollo, an artificial intelligence platform, for the development of autonomous military vehicles. If true, future technologies developed by China or Russia will heavily leverage the cloud to support military actions.
Peeking into Pandora’s Box
Developing programs that mimic human behaviors such as reasoning or judgement is a non-trivial matter. Hard-coding general human-like behavior is also unrealistic, because the code required would quickly become unwieldy in accounting for countless combinations of events or changes in contextual factors. Artificial intelligence technologies overcome the shortcomings of hard-coding human-like behaviors through a variety of techniques, but they also trade increased performance for greater opacity. For instance, under the umbrella of machine learning, techniques such as deep learning (e.g., neural networks, convolutional neural networks, etc.) use feature extraction techniques composed of many neural layers fed millions of examples to train and adjust the algorithm’s parameters. Thus, natural language processing and self-driving cars are made possible, not through hard-coding explicit instructions, but through different types and combinations of deep learning methods.
Understanding the results of algorithms that ingest petabytes of data, however, presents a black-box problem for consumers of such system outputs. Algorithmic obfuscation and bias are well-documented in a wide variety of fields from finance to social welfare. More importantly, a “war-algorithm” is now defined by a Harvard Law School report “as any algorithm that is expressed in computer code, that is effectuated through a constructed system, and that is capable of operating in relation to armed conflict.” The march towards greater automation will further conflate the lines between human and machine decision-making where a greater number of outputs are inextricably tied to artificial intelligence and machine learning technologies. It is therefore easy to surmise a future where war-algorithms running on cloud technologies approximate mythological gods dueling for supremacy on behalf of their military masters. Yet, if peer adversaries are faced with similar dilemmas, how can the U.S. forces respond effectively?
The 21st Century Herodotus
It is claimed that one of the foremost Greek historians, Herodotus, gathered a great deal of material from the Oracle at Delphi. Having an intimate understanding of history and an openness to the forecasts of an oracle, Herodotus may provide the best analog for future warriors to emulate. A warrior who possesses an appreciation of historical warfare and is tech-savvy enough to understand the benefits and limitations of artificial intelligence and machine learning may become the best synthesis of the ideal centaur. Yet, approaches to educating the force on these technologies are largely non-existent.
The U.S. military cannot wait for these technologies to mature before educating and training its personnel. Personnel conversant in knowledge of their particular warfighting domains and artificial intelligence and machine learning technologies will likely provide the best interdisciplinary approach to harnessing the power of these future systems. While not exhaustive, the following recommendations are immediate actions the military services can take to begin shaping the future and cost little to nothing to implement:
Professional Military Education — Professional military education institutions are the flagship establishments for educating large portions of the force. Combined with their grounded education in historical warfare, these institutions can begin integrating programs of informed instruction on how artificial intelligence and machine learning technologies may fit into the future force. This would also help move discussions away from shallow understanding of artificial intelligence or viewing it as a panacea for every capability gap. Furthermore, expansion of wargames and military case studies that introduce challenging dilemmas at the intersection of technology and military art will go a long way towards cultivating the mindsets needed in the 21st century service member. Continued improvement of on-demand video/podcast interviews with technological leaders could further spur creative thinking. Lastly, the introduction of discussions based around science fiction novels could also change the course of how personnel view future warfare.
Democratize Artificial Intelligence and Machine Learning Development and Experimentation — Men and women on the front lines should be the leaders of driving innovations in this space. Recently released planning guidance by the Commandant of the U.S. Marine Corps also recognizes the importance of individuals with artificial intelligence skill sets. Therefore, leveraging today’s open-source environments, which offer free classes and downloadable software packages offer the opportunity to anyone desiring to explore artificial intelligence and machine learning technologies. The Department of Defense can encourage individual exploration of these technologies through sanitized data sets of real-world data for experimentation and training machine learning models, which can help further understanding the benefits and limitations of artificial intelligence. Providing a military-centric contest similar to Kaggle competitions that solicits development of the most effective machine learning and deep learning models humans can muster could create incentives for soldiers, marines, sailors, and airmen to begin acquiring familiarity with these technologies.
Collaboration and Diffusion of Ideas — Today’s best artificial intelligence and machine learning initiatives are likely not happening in a secret underground lab hidden in the Nevada desert. Rather, individuals, teams, and research labs are collaborating across open-source platforms like GitHub to create artificial intelligence and machine learning applications. Organizations can identify areas where they may see immediate benefits of artificial intelligence and machine learning. Higher level leadership could solicit for a small number of volunteers from across multiple sub-divisions to bring together diverse teams to work on a high-payoff problem. Participants could run a three-month sprint and debrief a project to their leadership and post results on National Defense University’s National Security Innovation Network.
The U.S. military will need to continue its exploration of artificial intelligence and machine learning technologies across warfighting domains and functions. The generals and flag officers of 2030 are field grade officers today. Developing a talent pool for overseeing the development of future technologies is an essential part of an artificial intelligence strategy. Placing these technologies into the hands of warfighters sooner will allow the force to develop a greater appreciation for the benefits and potential shortcomings of these technologies in order to better prepare them to lead in the future.
Scholars and researchers have identified the various parallels between humanity’s quest to develop artificial intelligence technologies and Greek mythologies. Yet, these metaphors, albeit not perfect, may help raise important questions if artificial intelligence is treated like a god. The intersection of artificial intelligence, ethics, and warfare will undoubtedly raise more questions as the technology matures. Technologies such as artificial intelligence should not distract the Department of Defense from pursuing activities and research foundational to the conduct of human warfare. For these reasons, continuing to develop people through rigorous training, education, and an appreciation for the limits of artificial intelligence, will be vital on future battlefields where adversaries mirror each other’s technologies.
The intersection of artificial intelligence, ethics, and warfare will undoubtedly raise more questions as the technology matures.
Artificial intelligence and machine learning are likely to increase in importance as the technology scales and is woven into the fabric of one’s everyday life. Traffic navigation apps, digital assistants, autonomous vehicles, and the like, are becoming applications whose guidance is often followed without question. On the other hand, scalability in artificial intelligence and machine learning technologies could also lead from small scale implementation to applications of significant importance that become too big to fail. Here, Marine Corps Doctrine Publication 1 is instructive: “[t]here are two dangers with respect to equipment: the overreliance on technology and the failure to make the most of technological capabilities.” This makes heeding the warning of this doctrine problematic if artificial intelligence and machine learning are not closely monitored once they cross the threshold of becoming mission critical systems.
Ambiguity, opacity, bias, friction, and uncertainty have always been part and parcel of warfare in general and may become accepted realities of warfare driven by artificial intelligence. Nevertheless, the U.S. military cannot abandon core tenets of its development of human qualities that have provided the foundation for past and current successes. By advocating a dual approach of harnessing the benefits of artificial intelligence and machine learning technologies, while protecting a warrior’s ethos, the services can reap the benefits of developing 21st century centaurs.
Scott A. Humr is a United States Marine Corps officer. The views expressed are the author’s alone and do not reflect the official position of the U.S. Marine Corps, the Department of the Navy, the Department of Defense, or the U.S. Government.
This article appeared originally at Strategy Bridge.