Tag Archives: artificial intelligence

A Roadmap to Successful Sonar AI

Emerging Technologies Topic Week

By LT Andrew Pfau

Even as the private sector and academia have made rapid progress in the field of Artificial Intelligence (AI) and Machine Learning (ML), the Department of Defense (DoD) remains hamstrung by significant technical and policy challenges. Only a fraction of this civilian-driven progress can be applied to the AI and ML models and systems needed by the DoD; the uniquely military operational environments and modes of employment create unique development challenges for these potentially dominant systems. In order for ML systems to be successful once fielded, these issues must be considered now. The problems of dataset curation, data scarcity, updating models, and trust between humans and machines will challenge engineers in their efforts to create accurate, reliable, and relevant AI/ML systems.

Recent studies recognize these structural challenges. A GAO report found that only 38 percent of private sector research and development projects were aligned with DoD needs, while only 12 percent of projects could be categorized as AI or autonomy research.1 The National Security Commission on Artificial Intelligence’s Final Report also recognizes this gap, recommending more federal R&D funding for areas critical to advance technology, especially those that may not receive private sector investment.2 The sea services face particular challenges in adopting AI/ML technologies to their domains because private sector interest and investment in AI and autonomy at sea has been especially limited. One particular area that needs Navy-specific investment is that of ML systems for passive sonar systems, though the approach certainly has application to other ML systems.

Why Sonar is in Particular Need of Investment

Passive sonar systems are a critical component on many naval platforms today. Passive sonar listens for sounds emitted by ships or submarines and is the preferred tool of anti-submarine warfare, particularly for localizing and tracking targets. In contrast to active sonar, no signal is emitted, making it more covert and the method of choice for submarines to locate other vessels at sea. Passive sonar systems are used across the Navy in submarine, surface, and naval air assets, and in constant use during peace and war to locate and track adversary submarines. Because of this widespread use, any ML model for passive sonar systems would have a significant impact across the fleet and use on both manned and unmanned platforms. These models could easily integrate into traditional manned platforms to ease the cognitive load on human operators. They could also increase the autonomy of unmanned platforms, either surfaced or submerged, by giving these platforms the same abilities that manned platforms have to detect, track, and classify targets in passive sonar data.

Passive sonar, unlike technologies such as radar or LIDAR, lacks the dual use appeal that would spur high levels of private sector investment. While radar systems are used across the military and private sector for ground, naval, air, and space platforms, and active sonar has lucrative applications in the oil and gas industry, passive sonar is used almost exclusively by naval assets. This lack of incentive to invest in ML technologies related to sonar systems epitomizes the gap referred to by the NSC AI report. Recently, NORTHCOM has tested AI/ML systems to search through radar data for targets, a project that has received interest and participation from all 11 combatant commands and the DoD as a whole.3 Due to its niche uses, however, passive sonar ML systems cannot match this level of department wide investment and so demands strong advocacy within the Navy.

Dataset Curation

Artificial Intelligence and Machine Learning are often conflated and used interchangeably. Artificial Intelligence refers a field of computer science interested in creating machines that can behave with human-like abilities and can make decisions based on input data. In contrast, Machine Learning, a subset of the AI filed, refers to computer programs and algorithms that learn from repeated exposure to many examples, often millions, instead of operating based on explicit rules programmed by humans.4 The focus in this article is on topics specific to ML models and systems, which will be included as parts in a larger AI or autonomous system. For example, an ML model could classify ships from passive sonar data, this model would then feed information about those ships into an AI system that operates an Unmanned Underwater Vehicle (UUV). The AI would make decisions about how to steer the UUV based on data from the sonar ML model in addition to information about mission objectives, navigation, and other data.

Machine learning models must train on large volumes of data to produce accurate predictions. This data must be collected, labeled, and prepared for processing by the model. Data curation is a labor- and time-intensive task that is often viewed as an extra cost on ML projects since it must occur before any model can be trained, but this process should be seen as an integral part of ML model success. Researchers recently found that one of the most commonly used datasets in computer vision research, ImageNet, has approximately 6 percent of their images mislabeled 5. Another dataset, QuickDraw, had 10 percent of images mislabeled. Once the errors were corrected, model performance on the ImageNet dataset improved by 6 percent over a model trained on the original, uncorrected, dataset.5

For academic researchers, where the stakes of an error in a model are relatively low, this could be called a nuisance. However, ML models deployed on warships face greater consequences than those in research labs. A similar error, of 6 percent, in an ML model to classify warships would be far more consequential. The time and labor costs needed to correctly label data for use in ML model training needs to be factored into ML projects early. In order to make the creation of these datasets cost effective, automatic methods will be required to label data, and methods of expert human verification must ensure quality. Once a large enough dataset has been built up, costs will decrease. However, new data will still have to be continuously added to training datasets to ensure up to date examples are present in the training of models.

A passive acoustic dataset is much more than an audio recording: Where and when the data is collected, along with many other discrete factors, are also important and should be integrated into the dataset. Sonar data collected in one part of the ocean, or during a particular time of year, could be very different than other parts of the ocean or the same point in the ocean at a different time of year. Both the types of vessels encountered and the ocean environment will vary. Researchers at Brigham Young University demonstrated how variations in sound speed profiles can affect machine learning systems that operate on underwater acoustic data. They showed the effects of changing environmental conditions when attempting to classify seabed bottom type from a moving sound source, with variations in the ability of their ML model to provide correct classifications by up to 20 percent.6 Collecting data from all possible operating environments, at various times of the year, and labeling them appropriately will be critical to building robust datasets from which accurate ML models can be trained. Metadata, in the form of environmental conditions, sensor performance, sound propagation, and more must be incorporated during the data collection process. Engineers and researchers will be able to analyze metadata to understand where the data came from and what sensor or environmental conditions could be underrepresented or completely missing.

These challenges must be overcome in a cost-effective way to build datasets representative of real world operating environments and conditions.

Data Scarcity

Another challenge in the field of ML that has salience for sonar data are the challenges associated with very small, but important datasets. For an academic researcher, data scarcity may come about due to the prohibitive cost of experiments or rarity of events to collect data on, such as astronomical observations. For the DoN, these same challenges will occur in addition to DoN specific challenges. Unlike academia or the private sectors, stringent restrictions on classified data will limit who can use this data to train and develop models. How will an ML model be trained to recognize an adversary’s newest ship when there are only a few minutes of acoustic recording? Since machine learning models require large quantities of data, traditional training methods will not work or result in less effective models.

Data augmentation, replicating and modifying original data may be one answer to this problem. In computer vision research, data is augmented by rotating, flipping, or changing the color balance of an image. Since a car is still a car, even if the image of the car is rotated or inverted, a model will learn to recognize a car from many angles and in many environments. In acoustics research, data is augmented by adding in other sounds or changing the time scale or pitch of the original audio. From a few initial examples, a much larger dataset to train on can be created. However, these methods have not been extensively researched on passive sonar data. It is still unknown which methods of data augmentation will produce the best results for sonar models, and which could produce worse models. Further research into the best methods for data augmentation for underwater acoustics is required.

Another method used to generate training data is the use of models to create synthetic data. This method is used to create datasets to train voice recognition models. By using physical models, audio recordings can be simulated in rooms of various dimensions and materials, instead of trying to make recordings in every possible room configuration. Generating synthetic data for underwater audio is not as simple and will require more complex models and more compute power than models used for voice recognition. Researchers have experimented with generated synthetic underwater sounds using the ORCA sound propagation model.6 However, this research only simulated five discrete frequencies used in seabed classification work. A ship model for passive sonar data will require more frequencies, both discrete and broadband, to be simulated in order to produce synthetic acoustic data with enough fidelity to use in model training. The generation of realistic synthetic data will give system designers the ability to add targets with very few examples to a dataset.

The ability to augment existing data and create new data from synthetic models will create larger and more diverse datasets, leading to more robust and accurate ML models.

Building Trust between Humans and Machines

Machine learning models are good at telling a human what they know, which comes from the data they were trained on. They are not good at telling humans that they do not recognize an input or have never seen anything like it in training. This will be an issue if human operators are to develop trust in the ML models they will use. Telling an operator that it does not know, or the degree of confidence a model has in its answer, will be vital to building reliable human-machine teams. One method to building models with the ability to tell human operators that a sample is unknown is the use of Bayesian Neural Networks. Bayesian models can tell an operator how confident they are in a classification and even when the model does not know the answer. This falls under the field of explainable AI, AI systems that can tell a human how the system arrived at the classification or decision that is produced. In order to build trust between human operators and ML systems, a human will need some insight into how and why an ML system arrived at its output.

Ships at sea will encounter new ships, or ships that were not part of the model’s original training dataset. This will be a problem early in the use of these models, as datasets will initially be small and grow with the collection of more data. These models cannot fail easily and quickly, they must be able to distinguish between what is known and what is unknown. The DoN must consider how human operators will interact with these ML models at sea, not just model performance.

Model Updates

To build a great ML system, the models will have to be updated. New data will be collected and added to the training dataset to re-train the model so that it stays relevant. In these models, only certain model parameters are updated, not the design or structure of the model. These updates, like any other digital file can be measured in bytes. An important question for system designers to consider is how these updates will be distributed to fleet units and how often. One established model for this is the Acoustic- Rapid COTS Insertion (ARCI) program used in the US Navy’s Submarine Force. In the ARCI program, new hardware and software for sonar and fire control is built, tested, and deployed on a regular, two-year cycle.7 But two years may be too infrequent for ML systems that are capable of incorporating new data and models rapidly. The software industry employs a system of continuous deployment, in which engineers can push the latest model updates to their cloud-based systems instantly. This may work for some fleet units that have the network bandwidth to support over the air updates or that can return to base for physical transfer. Recognizing this gap, the Navy is currently seeking a system that can simultaneously refuel and transfer data, up to 2 terabytes, from a USV.8 This research proposal highlights the large volume of data will need to be moved, both on and off unmanned vessels. Other units, particularly submarines and UUVs, have far less communications bandwidth. If over-the-air updates to submarines or UUVs are desired, then more restrictions will be placed on model sizes to accommodate limited bandwidth. If models cannot be made small enough, updates will have to be brought to a unit in port and updated from a hard drive or other physical device.

Creating a good system for when and how to update these models will drive other system requirements. Engineers will need these requirements, such as size limitations on the model, ingestible data type, frequency of updates needed by the fleet, and how new data will be incorporated into model training before they start designing ML systems.

Conclusion

As recommended in the NSC AI report, the DoN must be ready to invest in technologies that are critical to future AI systems, but that are currently lacking in private sector interest. ML models for passive sonar, lacking both dual use appeal and broad uses across the DoD, clearly fits into this need. Specific investment is required to address several problems facing sonar ML systems, including dataset curation, data scarcity, model updates, and building trust between operators and systems. These challenges will require a combination of technical and policy solutions to solve them, and they must be solved in order to create successful ML systems. Addressing these challenges now, while projects are in a nascent stage, will lead to the development of more robust systems. These sonar ML systems will be a critical tool across a manned and unmanned fleet in anti-submarine warfare and the hunt for near-peer adversary submarines.

Lieutenant Andrew Pfau, USN, is a submariner serving as an instructor at the U.S. Naval Academy. He is a graduate of the Naval Postgraduate School and a recipient of the Rear Admiral Grace Murray Hopper Computer Science Award. The views and opinions expressed here are his own.

Endnotes

1. DiNapoli, T. J. (2020). Opportunities to Better Integrate Industry Independent Research and Development into DOD Planning. (GAO-20-578). Government Accountability Office.

2. National Security Commission on Artificial Intelligence (2021), Final Report.

3. Hitchens, T. (2021, July 15) NORTHCOM Head To Press DoD Leaders For AI Tools, Breaking Defense, https://breakingdefense.com/2021/07/exclusive-northcom-head-to-press-dod-leaders-for-ai-tools/

4. Denning, P., Lewis, T. Intelligence May Not be Computable. American Scientist. Nov-Dec 2019. http://denninginstitute.com/pjd/PUBS/amsci-2019-ai-hierachy.pdf

5. Hao, K. (2021, April 1) Error-riddled data sets are warping our sense of how good AI really is. MIT Technology Review. https://www.technologyreview.com/2021/04/01/1021619/ai-data-errors-warp-machine-learning-progress/

6. Neilsen et al (2021). Learning location and seabed type from a moving mid-frequency source. Journal of the Acoustical Society of America. (149). 692-705. https://doi.org/10.1121/10.0003361

7. DeLuca, P., Predd, J. B., Nixon, M., Blickstein, I., Button, R. W., Kallimani J. G., and Tierney, S. (2013) Lessons Learned from ARCI and SSDS in Assessing Aegis Program Transition to an Open-Architecture Model, (pp 79-84) RAND Corperation, https://www.jstor.org/stable/pdf/10.7249/j.ctt5hhsmj.15.pdf

8. Office of Naval Research, Automated Offboard Refueling and Data Transfer for Unmanned Surface Vehicles, BAA Announcement # N00014-16-S-BA09, https://www.globalsecurity.org/military/systems/ship/systems/oradts.htm

Featured Image: Sonar Technician (Surface) Seaman Daniel Kline performs passive acoustic analysis in the sonar control room aboard the guided-missile destroyer USS Ramage (DDG 61). (U.S. Navy photo by Mass Communication Specialist 2nd Class Jacob D. Moore/Released)

Winning The AI-Enabled War-at-Sea

By Dr. Peter Layton

Artificial intelligence (AI) technology is suddenly important to military forces. Not yet an arms race, today’s competition is more in terms of an experimentation race with many AI systems being tested and new research centers established. There may be a considerable first-mover advantage to the country that first understands AI adequately enough to change its existing human-centered force structures and embrace AI warfighting.

In a new Joint Studies Paper, I explore sea, land and air operational concepts appropriate to fighting near-to-medium term future AI-enabled wars. With much of the underlying narrow AI technology already developed in the commercial sector, this is less of a speculative exercise than might be assumed. Moreover, the contemporary AI’s general-purpose nature means its initial employment will be within existing operational level constructs, not wholly new ones.

Here, the focus is the sea domain. The operational concepts mooted are simply meant to stimulate thought about the future and how to prepare for it. In being so aimed, the concepts are deliberately constrained; crucially they are not joint or combined. In all this, it is important to remember that AI enlivens other technologies. AI is not a stand-alone actor, rather it works in the combination with numerous other digital technologies. It provides a form of cognition to these.

AI Overview

In the near-to-medium term, AI’s principal attraction is its ability to quickly identify patterns and detect items hidden within very large data troves. The principal consequence of this is that AI will make it much easier to detect, localize and identity objects across the battlespace. Hiding will become increasingly difficult. However, AI is not perfect. It has well known problems in being able to be fooled, in being brittle, being unable to transfer knowledge gained in one task to another and being dependent on data.

AI’s warfighting principal utility then becomes ‘find and fool’. AI with its machine learning is excellent at finding items hidden within a high clutter background. In this role AI is better than humans and tremendously faster. On the other hand, AI can be fooled through various means. AI’s great finding capabilities lack robustness.

A broad generic overview is useful to set the scene. The ‘find’ starting point is placing a large number of low cost Internet of Things (IoT) sensors in the optimum land, sea, air, space and cyber locations in the areas across which hostile forces may transit. From these sensors, a deep understanding can be gained of the undersea terrain, sea conditions, physical environment and local virtual milieu. Having this background data accelerates AI’s detection of any changes and, in particular, of the movement of military forces across it.

The fixed and mobile IoT edge-computing sensors are connected into a robust cloud to reliably feed data back into remote command support systems. The command system’s well-trained AI could then very rapidly filter out the important information from the background clutter. Using this, AI can then forecast adversary actions and predict optimum own force employment and its combat effectiveness. Hostile forces geolocated by AI can, after approval by human commanders, be quickly engaged using indirect fire including long-range missiles. Such an approach can engage close or deep targets; the key issues being data on the targets and the availability of suitable range firepower. The result is that the defended area quickly becomes a no-go zone.

To support the ‘fool’ function, Uncrewed Vehicles (UV) could be deployed across the battlespace equipped with a variety of electronic systems suitable for the Counter Intelligence Surveillance And Reconnaissance And Targeting (C-ISRT) task. The intent is to defeat the adversary’s AI ‘find’ capabilities. Made mobile through AI, these UVs will be harder for an enemy to destroy than fixed jammers would be. Moreover, mobile UVs can be risked and sent close in to approaching hostile forces to maximize jamming effectiveness. Such vehicles could also play a key role in deception, creating a false and misleading impression of the battlefield to the adversary. Imagine a battlespace where there are a thousand ‘valid’ targets, only a few of which are real.

A War-at-Sea Defense Concept

Defense is the more difficult tactical problem during a war-at-sea. Its intent is solely to gain tactical time for an effective attack or counterattack. Wayne Hughes goes as far in his seminal work to declare that: “All fleet operations based on defensive tactics…are conceptually deficient.”1  The AI-enabled battlefield may soften this assertion.

Accurately determining where hostile ships are in the vast ocean battlefields has traditionally been difficult. A great constant of such reconnaissance is that there never seems to be enough. However, against this, a great trend since the early 20th century is that maritime surveillance and reconnaissance technology is steadily improving. The focus is now not on collecting information but on improving the processing of the large troves of surveillance and reconnaissance data collected.2 Finding the warship ‘needle’ in the sea ‘haystack’ is becoming easier. 

The earlier generic ‘find’ concept envisaged a large distributed IoT sensor field. Such a concept is becoming possible in the maritime domain given AI and associated technology developments.

DARPA’s Ocean of Things (OoT) program aims to achieve maritime situational awareness over large ocean areas through deploying thousands of small, low-cost floats that form a distributed sensor network. Each smart float will have a suite of commercially available sensors to collect environmental and activity data; the later function involves automatically detecting, tracking and identifying nearby ships and – potentially – close aircraft traffic. The floats use edge processing with detection algorithms and then transmit the semi-processed data periodically via the Iridium satellite constellation to a cloud network for on-shore storage. AI machine learning then combs through this sparse data in real time to uncover hidden insights. The floats are environmentally friendly, have a life of around a year and in buys of 50,000 have a unit cost of about US$500 each. DARPA’s OoT shows what is feasible using AI.

In addition to floats, there are numerous other low-cost AI-enabled mobile devices that could noticeably expand maritime situational awareness including: the EMILY Hurricane Trackers, Ocean Aero Intelligent Autonomous Marine Vehicles, Seaglider Autonomous Underwater Vehicles, Liquid Robotics Wave Gliders and Australia’s Ocius Technology Bluebottles.

In addition to mobile low-cost autonomous devices plying the seas there is an increasing number of smallsats being launched by governments and commercial companies into low earth orbit to form large constellations. Most of these will use AI and edge computing; some will have sensors able to detect naval vessels visually or electronically.

All this data from new sources can be combined with that from the existing large array of traditional maritime surveillance systems. The latest system into service is the long-endurance MQ-4C Triton uncrewed aerial vehicle with detection capabilities able to be enhanced through retrofitting AI. The next advance may be the USN’s proposed 8000km range, AI-enabled Medium Unmanned Surface Vessel (MUSV) which could cruise autonomously at sea for two months with a surveillance payload.

With so many current and emerging maritime surveillance systems, the idea of a digital ocean is becoming practical. This concept envisages the data from thousands of persistent and mobile sensors being processed by AI, analyzed though machine learning and then fused into a detailed ocean-spanning three-dimensional comprehensive picture. Oceans remain large expanses making this a difficult challenge. However, a detailed near-real time digital model of smaller spaces such as enclosed waters like the South China Sea, national littoral zones or limited ocean areas of specific import appears practical using current and near-term technology.

Being able to create a digital ocean model may prove revolutionary. William Williamson of the USN Naval Postgraduate School declares: “On the ‘observable ocean’, the Navy must assume that every combatant will be trackable, with position updates occurring many times per day. …the Navy will have lost the advantages of invisibility, uncertainty, and surprise. …Vessels will be observable in port…[with] the time of departure known to within hours or even minutes. This is true for submarines as well as for surface ships.”3

This means that in a future major conflict, the default assessment by each warship’s captain might be that the adversary probably knows the ship’s location. Defense then moves from being “conceptually deficient” to being the foundation of all naval tactics in an AI-enabled battlespace. The emerging AI-enabled maritime surveillance system of systems will potentially radically change traditional war-at-sea thinking. The ‘attack effectively first’ mantra may need to be rewritten to ‘defend effectively first.’

The digital, ‘observable ocean’ will ensure warships are aware of approaching hostile warships and a consequent increasing risk of attack. In this addressing this, three broad alternative ways for the point defense of a naval task group might be considered.

Firstly, warships might cluster together, so as to concentrate their defensive capabilities and avoid any single ship being overwhelmed by a large multi-axis, multi-missile attack. In this, AI-enabled ship-borne radars and sensors will be able to better track incoming missiles amongst the background clutter. Moreover, AI-enabled command systems will be able to much more rapidly prioritize and undertake missile engagements. In addition, nearby AI-enabled uncrewed surface vessels may switch on active illuminator radars, allowing crewed surface combatants to receive reflections to create fire control-quality tracks. The speed and complexity of the attacks will probably mean that human-on-the-loop is the generally preferred AI-enabled ship weapon system control, switching to human-out-of-the-loop as numbers of incoming missiles rise or hypersonic missiles are faced.

Secondly, instead of clustering, warships might scatter so that an attack against one will not endanger others. Crucially, modern technology now allows dispersed ships to fight together as a single package. The ‘distributed lethality’ concept envisages distant warships sharing precise radar tracking data across a digital network, although there are issues of data latency that limit how far apart the ships sharing data for this purpose can be. An important driver of the ‘distributed lethality’ concept is to make adversary targeting more difficult. With the digital ocean, this driver may be becoming moot.

Thirdly, the defense in depth construct offers new potential through becoming AI-enabled, particularly when defending against submarines although the basic ideas also have value against surface warship threats. In areas submarines may transit through, stationary relocatable sensors like the USN’s Transformational Reliable Acoustic Path System could be employed backed up by unpowered, long endurance gliders towing passive arrays. These passive sonars would use automated target recognition algorithms supported by AI machine learning to identify specific underwater or surface contacts.

Closer to the friendly fleet, autonomous MUSVs could use low-frequency active variable depth sonars supplemented by medium-sized uncrewed underwater vehicles (UUV) with passive sonar arrays. Surface warships or the MUSVs could further deploy small UUVs carrying active multistatic acoustic coherent sensors already fielded in expendable sonobuoys. Warships could employ passive sonars to avoid counter-detection and take advantage of multistatic returns from the active variable depth sonars deployed by MUSVs.

Fool Function. The “digital ocean” significantly increases the importance of deception and confusion operations. This ‘fool’ function of AI may become as vital as the ‘find’ function, especially in the defense. In the war-at-sea, the multiple AI-enabled systems deployed across the battlespace offer numerous possibilities for fooling the adversary.

Deception involves reinforcing the perceptions or expectations of an adversary commander and then doing something else. In this, multiple false cues will need seeding as some clues will be missed by the adversary and having more than one will only add to the deception’s credibility. For example, a number of uncrewed surface vessels could set sail as the warship leaves port, all actively transmitting a noisy facsimile of the warships electronic or acoustic signature. The digital ocean may then suggest to the commander multiple identical warships are at sea, creating some uncertainty as to which is real or not.

In terms of confusion, the intent might be not to avoid detection as this might be very difficult but instead prevent an adversary from classifying vessels detected as warships or identifying them as a specific class of warship. This might be done using some of the large array of AI-enabled floaters, gliders, autonomous devices, underwater vehicles and uncrewed surface vessels to considerably confuse the digital ocean picture. The aim would be to change the empty oceans – or at least the operational area – into a seemingly crowded, cluttered, confusing environment where detecting and tracking the real sought-after warships was problematic and at best fleeting. If AI can find targets, AI can also obscure them.

A War-at-Sea Offense Concept

In a conflict where both sides are employing AI-enabled ‘fool’ systems, targeting adversary warships may become problematic. The ‘attack effectively first’ mantra may evolve to simply ‘attack effectively.’ Missiles that miss represent a significant loss of the task group’s or fleet’s net combat power, and take a considerable time to be replaced. Several alternatives may be viable.

In a coordinated attack, the offence might use a mix of crewed and uncrewed vessels. One option is to use three ship types: a large, well-defended crewed ship that carries considerable numbers of various types of long-range missiles but which remains remote to the high-threat areas; a smaller crewed warship pushed forward into the area where adversary ships are believed to be both for reconnaissance and to provide targeting for the larger ship’s long-range missiles; and an uncrewed stealthy ship operating still further forward in the highest risk area primarily collecting crucial time-sensitive intelligence and passing this back through the smaller crewed warship onto the larger ship in the rear.

The intermediate small crewed vessel can employ elevated or tethered systems and uncrewed communications relay vehicles to receive the information from the forward uncrewed vessel and act as a robust gateway to the fleet tactical grid using resilient communications systems and networks. Moreover, the intermediate smaller crewed vessel in being closer to the uncrewed vessel will be able to control it as the tactical situation requires and, if the context changes, adjust the uncrewed vessel’s mission.

This intermediate ship will probably also have small numbers of missiles available to use in extremis if the backward link to the larger missile ship fails. Assuming communications to all elements of the force will be available in all situations may be unwise. The group of three ships should be network enabled, not network dependent, and this could be achieved by allowing the intermediate ship to be capable of limited independent action.

The coordinated attack option is not a variant of the distributed lethality concept noted earlier. The data being passed from the stealthy uncrewed ship and the intermediate crewed vessel is targeting, not fire control, quality data. The coordinated attack option has only loose integration that is both less technically demanding and more appropriate to operations in an intense electronic warfare environment.

An alternative concept is to have a large crewed vessel at the center of a networked constellation of small and medium-sized uncrewed air, surface and subsurface systems. A large ship offers potential advantages in being able to incorporate advanced power generation to support emerging defensive systems like high energy lasers or rail guns. In this, the large crewed ship would need good survivability features, suitable defensive systems, an excellent command and control system to operate its multitude of diverse uncrewed systems and a high bandwidth communication system linking back to shore-based facilities and data storage services.

The crewed ship could employ mosaic warfare techniques to set up extended kinetic and non-kinetic kill webs through the uncrewed systems to reach the adversary warships. The ship’s combat power is not then in the crewed vessel but principally in its uncrewed systems with their varying levels of autonomy, AI application and edge computing.

The large ship and its associated constellation would effectively be a naval version of the Soviet reconnaissance-strike complex.  An AI-enabled war at sea then might involve dueling constellations, each seeking relative advantage.

Conclusion

The AI-enabled battlespace creates a different war-at-sea. Most obvious are the autonomous systems and vessels made possible by AI and edge computing. The bigger change though may be to finally take the steady scouting improvements of the last 100 years or so to their final conclusion. The age of AI, machine learning, big data, IoT and cloud computing appear set to create the “observable ocean.” From combining these technologies, near-real digital models of the ocean environment can be made that highlight the man-made artefacts present.

The digital ocean means warships could become the prey as much as the hunters. Such a perspective brings a shift in thinking about what the capital ship of the future might be. A recent study noted: “Navy’s next capital ship will not be a ship. It will be the Network of Humans and Machines, the Navy’s new center of gravity, embodying a superior source of combat power.” Tomorrow’s capital ship looks set to be the human-machine teams operating on an AI-enabled battlefield.

Dr. Peter Layton is a Visiting Fellow at the Griffith Asia Institute, Griffith University and an Associate Fellow at the Royal United Services Institute. He has extensive aviation and defense experience and, for his work at the Pentagon on force structure matters, was awarded the US Secretary of Defense’s Exceptional Public Service Medal. He has a doctorate from the University of New South Wales on grand strategy and has taught on the topic at the Eisenhower School. His research interests include grand strategy, national security policies particularly relating to middle powers, defense force structure concepts and the impacts of emerging technology. The author of ‘Grand Strategy’, his posts, articles and papers may be read at: https://peterlayton.academia.edu/research.

Endnotes

1. Wayne P. Hughes and Robert Girrier, Fleet tactics and naval operations, 3rd edn., (Annapolis: Naval Institute Press, 2018), p. 33.

2. Ibid., pp.132, 198.

3. William Williamson, ‘From Battleship to Chess’, USNI Proceedings, Vol. 146/7/1,409, July 2020, https://www.usni.org/magazines/proceedings/2020/july/battleship-chess

Featured image: Graphic highlighting Fleet Cyber Command Watch Floor of the U.S. Navy. (U.S. Navy graphic by Oliver Elijah Wood and PO2 William Sykes/Released)

Virtual Training: Preparing Future Naval Officers for 21st Century Warfare

By Joseph Bunyard

Introduction

“[We must] embrace the urgency of the moment: our maritime supremacy is being challenged.” —CNO NAVPLAN 2021

The fundamental character of war is changing.1 Distributed networks, next generation threats, and artificial intelligence will change “the face of conflict” by compressing and accelerating the Observe, Orient, Decide, Act (OODA) loop, streamlining the closure of kill chains.2 American security depends on the Navy’s ability to control the seas and project power ashore.3 Preparing future naval officers for 21st century warfare must begin at the US Naval Academy (USNA), where Virtual Training Environments (VTEs) could provide education and training opportunities once exclusive to the Fleet.4

21st century warfare requires data producers and smart data consumers. Although the Department of Defense recognizes the need for an “AI ready force,” the 2018 National Defense Strategy claims that professional military education “has stagnated at the expense of lethality and ingenuity.”5 To address this charge, the Navy’s 2020 Education for Seapower Strategy calls for the creation of a “continuum of learning” through the Naval University System.6 While the Naval Postgraduate School conducts innovative technical research—and the Naval War College endows senior leaders with a strategic outlook on the future of warfare—the US Naval Academy does not feature AI, unmanned systems, tactics, or strategy in its core curriculum.7

Figure 1 – Aviation Officer Career Progression. Above: aviation officers require 2.5 years of training before deployment. 8

New technology often means new qualification requirements for junior officers. Added training extends the length of time before an officer is ready to deploy, a worrying trend at which Type Commanders are taking aim (see Figure 1).9 VTEs could offer Midshipmen exposure to the naval applications of disruptive technologies, the chance to accomplish existing Fleet training prior to commissioning, and Artificial Intelligence (AI)/ Machine Learning (ML) tools that they could take to the Fleet. To realize these objectives, the Naval Academy must leverage three types of VTEs—low-cost, commercial-off-the-shelf (COTS), and Fleet-integrated—to expand training opportunities and reinforce its core curriculum.

E-learning in the COVID-19 era provides the Naval Academy a chance to update its operating system (OS). Instead of using new media, such as Zoom, to present the same PowerPoints Midshipmen would receive in-person, USNA should update its curriculum to take advantage of VTEs with proven training and educational outcomes. Incorporating new media into existing curricula requires an OS update that expands USNA’s “leadership laboratory” into a 21st century warfare laboratory, where smart data producers and consumers are forged. 10

Integrating Low-Cost Virtual Training Environments (VTEs)

“To maintain naval power in an era of great power competition and technological change, the Navy and Marine Corps need to strengthen and expand their educational efforts.”—Education for Seapower Strategy 2020

The Navy and Marine Corps increasingly rely on VTEs to “expand watch team proficiency and combat readiness” across the Fleet.11 Unlike traditional simulators, virtual reality trainers are highly mobile and often rely on commercial-off-the-shelf (COTS) hardware. The Chief of Naval Air Training’s Project Avenger simulator, for example, uses gaming computers and virtual reality headsets to qualify students for solo flights in half of the traditional number of flight hours.12 The Marine Corps’ tactical decision kits use similar technology to train infantry battalions on weapon systems and tactics.13 Mixed reality glasses, which overlay a user’s vision with digital information, help crews across the Fleet complete complex maintenance.14

Expanding access to existing virtual reality trainers at the Naval Academy could enable Midshipmen to complete portions of Naval Introductory Flight Evaluation (NIFE), The Basic School (TBS), and Basic Division Officer Course (BDOC) syllabi prior to commissioning. “Future multi-domain combat will be so complex and long-ranged that the military will rely heavily on simulations to train for it.”15 More access to VTE trainers means more familiarization with the technology and interfaces that junior officers are increasingly likely to encounter in the Fleet.

Figure 2 – A Project Avenger Simulator. U.S. Navy photo. 16

Accessing the Navy Continuous Training Environment (NCTE)

“Winning in contested seas also means fielding and equipping teams that are masters of all-domain fleet operations.” —CNO NAVPLAN 2021

VTEs allow users to conceptualize next generation threats. While the Naval Academy provides Midshipmen the technical foundation to understand Anti-Access/ Area-Denial (A2/AD) bubbles and contested communications zones, it offers few means for Midshipmen to visualize these abstract threats in an operational context.17 NAVAIR’s Joint Simulation Environment (JSE) and INDOPACCOM’s Pacific Multi-Domain Training and Experimentation Capability simulate next generation threats for operations analysis and platform research design testing and evaluation (RDT&E).18 The Navy Continuous Training Environment (NCTE) enables cross-platform integration of these platforms, and many more, which allows warfighters around the world to take part in scalable multi-domain battle problems.19

Figure 3 – NAVAIR’s JSE 20

To meet the Fleet’s growing need for diversified data, the Navy should leverage the informed and available, yet inexperienced, potential of the Academy’s more than 4,000 Midshipmen. Providing the Naval Academy with NCTE access could generate data for the Fleet and the operational context of classroom lessons for Midshipmen. Data is the new oil; improving predictive AI/ML models, concepts of operation, and training interfaces requires mass amounts of quality data from a range of problem-solving approaches.21 Installing an NCTE node in Hopper Hall’s new Sensitive Compartmented Information Facilities (SCIFs) would not only allow Midshipmen to observe Fleet training events but also to perform their own operations analysis on platforms, capabilities, and strategies developed during their capstone research.22

Leveraging Commercial-Off-The-Shelf (COTS) VTEs

“Advances in artificial intelligence and machine learning have increased the importance of achieving decision superiority in combat.” —CNO NAVPLAN 2021

For the cost of a video game, the Naval Academy could use the same software as defense industry leaders to improve the decision-making ability of Midshipmen, reinforce classroom concepts, and introduce next generation threats and platforms. The Defense Advanced Research Projects Agency (DARPA) uses popular videogames like Command: Modern Operations ($79.99 on Steam) to search for “asymmetrical conditions” within “hyper-realistic theater-wide combat simulators” that could be exploited in real-world scenarios.23 Many titles offer open Application Programming Interfaces (APIs) that allow users to change the decision-making logic of AI opponents and load custom platforms and capabilities into the game, such as squadrons of future unmanned systems.24 Modern concepts of operation—like Expeditionary Advanced Basing Operations and Joint All-Domain Command Control—often undergo “virtual sea trials” in such simulations.25

Figure 4 – Simulated Theater-Level Conflict in the South China Sea

The user-friendly, scalable, and unclassified nature of wargame simulators like Command: Modern Operations make them suitable for inter-academy use. Allies such as the United Kingdom already use commercial titles to host “Fight Clubs” among military and civilian personnel across all roles and ranks of their armed forces.26 By leveraging its cadre of foreign exchange officers and multilateral relationships, the Naval Academy could form an international “fight club” in the style of the growing “e-sports” industry. Competing with and against international Midshipmen and officers would allow Naval Academy Midshipmen to forge relationships with allies and learn from their approaches to tactics, strategies, and decision-making across a variety of simulated scenarios.

COTS Artificial Intelligence (AI) & Machine Learning (ML) VTEs

“Adopting AI capabilities at speed and scale is essential to maintain military advantage.”—2020 Department of Defense AI Education Strategy

Virtual machines provide users with access to advanced AI and ML tools, as well as the computing power necessary to use them at scale, anywhere there is an internet connection.27 Maintaining the Navy’s military advantage requires an “AI ready force” of smart data producers and consumers.28 Applying AI to operations and processes across the Fleet will likely make open-source ML software the Excel of the future, requiring both smart data producers and consumers. Not every officer is an Excel “wizard,” but most understand how it works, the problems it can solve, and the type of data it needs to function. In order to build an “AI ready force” across all roles and ranks, the Naval Academy should join the growing field of leading research universities incorporating introductory AI and ML courses in their core curricula.29

Just as seamanship and navigation are the cornerstone of maritime competence, AI-literacy will be the core of digital competence. Incorporating AI and ML into the Naval Academy’s core curriculum would create smart data producers and consumers, accelerating the Fleet’s exposure to AI through the bottom up approach envisioned in the Department of Defense AI Education Strategy.30 According to a 2019 study by IBM, “model interoperability,” understanding how a model arrives at a given decision is the single factor that most influences users’ trust in AI.31 Naval Academy graduates literate in AI and ML could better lead enlisted sailors as increasingly complex systems join the Fleet.

Towards a 21st Century Warfare Laboratory

“Transforming our learning model for the 21st century will enable us to adapt and achieve decisive advantage in complex, rapidly changing operating environments.” —2020 Triservice Maritime Strategy 32

The Naval Academy must return to the warfighting mentality of its past.33 In 2007, the Naval Academy not only removed its only tactics and strategy course from the Midshipmen core curriculum, it stopped offering it altogether.34 Until recently, this decision signaled the end of a rich history of wargaming at USNA, which included Academy-wide games held at varying levels of classification.35 VTEs offer the Naval Academy an opportunity to reprioritize warfighting by providing the “ready, relevant learning” future naval officers will need to conduct 21st century warfare.36

New concepts of operation require learning and experimentation that 21st century warfare-literate junior officers could accelerate. The Navy and Marine Corps continue to outline ambitious plans that leverage AI, unmanned platforms, and next generation networks in new concepts of operation. Consequently, the Navy aims to equip sailors with “a high degree of confidence and skill operating alongside” unmanned platforms and AI by “the end of this decade.”37 Creating a true “learning continuum” to prepare the Fleet for the future of warfare must start at the US Naval Academy, where the COVID-19 distance-learning environment offers an opportunity for the Naval Academy to update its operating system using VTEs.

Ensign Bunyard is a 2020 graduate of the U.S. Naval Academy. Upon completing his Master’s in Information Technology Strategy at Carnegie Mellon University, he will report to Pensacola for training as a student naval aviator.

Endnotes

1. Grady, John, and Sam Sam Lagrone. “CJCS Milley: Character of War in Midst of Fundamental Change.” USNI News, December 4, 2020. https://news.usni.org/2020/12/04/cjcs-milley-character-of-war-in-midst-of-fundamental-change.
2. Kitchener, Roy, Brad Cooper, Paul Schlise, Thibaut Delloue, and Kyle Cregge. “What Got Us Here Won’t Get Us There.” U.S. Naval Institute, January 9, 2021. https://www.usni.org/magazines/proceedings/2021/january/what-got-us-here-wont-get-us-there.
3. Gilday, Mike M. CNO NAVPLAN 2021. Office of the Chief of Naval Operations. Accessed February 2, 2021. https://media.defense.gov/2021/Jan/11/2002562551/-1/-1/1/CNO%20NAVPLAN%202021%20-%20FINAL.PDF., 4.
4. Wilson, Clay. Network Centric Warfare: Background and Oversight Issues for Congress. CRS Report for Congress § (2005).
5. Mattis, Jim. “Summary of the 2018 National Defense Strategy.” Department of Defense Media. Office of the Secretary of Defense, n.d. Accessed February 2, 2021., 8.
6. Gilday, 4.
7. “USNA Core Curriculum.” The U.S. Naval Academy. Accessed February 2, 2021. https://www.usna.edu/Academics/Majors-and-Courses/Course-Requirements-Core.php.
8. Morris, Terry. “Promotion Boards Brief.” Navy Personnel Command. Accessed February 2, 2021. https://slideplayer.com/slide/11144308/.
9. Shelbourne, Mallory. “Navy Harnessing New Technology to Restructure Aviation Training.” USNI News, September 14, 2020. https://news.usni.org/2020/09/14/navy-harnessing-new-technology-to-restructure-aviation-training.
10. Miller, Christopher A. “The Influence of Midshipmen on Leadership of Morale at the United States Naval Academy.” Naval Post Graduate School Thesis. Naval Post Graduate School. Accessed February 2, 2021. https://apps.dtic.mil/dtic/tr/fulltext/u2/a462636.pdf.
11. Kitchener, Roy.
12. Freedburg, Sydney J. “Project Avenger: VR, Big Data Sharpen Navy Pilot Training.” Breaking Defense. Above the Law, December 4, 2020. https://breakingdefense.com/2020/12/project-avenger-vr-big-data-sharpen-navy-pilot-training/
13. Berger, David. “Tactical Decision Kit Distribution and Implementation.” MARADMIN. US Marine Corps. Accessed February 2, 2021. https://www.marines.mil/News/Messages/Messages-Display/Article/1176937/tactical-decision-kit-distribution-and-implementation/.
14. Fretty, Peter. “Augmented Reality Helps US Navy See Clearer.” Industry Week. Accessed February 2, 2021. https://www.industryweek.com/technology-and-iiot/article/21142049/us-navy-sees-augmented-reality.
15. Freedburg, Sydney J. “Navy, Marines Plan Big Wargames For Big Wars: Virtual Is Vital.” Breaking Defense. Above the Law, December 3, 2020. https://breakingdefense.com/2020/12/navy-marines-plan-big-wargames-for-big-wars-virtual-is-vital/.
16. Shelbourne, Mallory.
17. Gonzales, Matt. “Marine Corps to Build Innovative Wargaming Center.” United States Marine Corps Flagship, August 25, 2020. https://www.marines.mil/News/News-Display/Article/2323771/marine-corps-to-build-innovative-wargaming-center/.
18. Davidson, Philip S. “Statement of Admiral Philip S. Davidson, US Navy Commander, US Indo-Pacific Command Before the Senate Armed Services Committee on US Info-Pacific Command Posture 12 February 2019.” Senate Armed Services Committee, February 12, 2019. https://www.armed-services.senate.gov/imo/media/doc/Davidson_02-12-19.pdf.
19. “Joint Simulation Environment.” NAVAIR. Naval Air Warfare Center. Accessed February 2, 2021. https://www.navair.navy.mil/nawctsd/sites/g/files/jejdrs596/files/2018-11/2018-jse.pdf. Also, Squire, Peter. “Augmented Reality Efforts.” Office of Naval Research. Accessed February 2, 2021., 13.
20. “Joint Simulation Environment.”
21. Graham, Karen. “AI Systems Are ‘Only as Good as the Data We Put into Them’.” Digital Journal: A Global Digital Media Network, September 5, 2018. http://www.digitaljournal.com/tech-and-science/technology/a-i-systems-are-only-as-good-as-the-data-we-put-into-them/article/531246. Also, Nilekani, Nandan. “Data to the People.” Foreign Affairs. Council on Foreign Relations, July 29, 2020. https://www.foreignaffairs.com/articles/asia/2018-08-13/data-people.
22. Tortora, Paul. “Center for Cyber Security Studies – 2018-2019 Stewardship Report.” Cyber Studies, March 14, 2020. http://1970.usnaclasses.com/Classprojects/Center%20for%20Cyber%20Studies.html.
23. Atherton, Kelsey. “DARPA Wants Wargame AI To Never Fight Fair.” Breaking Defense. Above the Law, August 18, 2020. https://breakingdefense.com/2020/08/darpa-wants-wargame-ai-to-never-fight-fair/. Also, “Command: Modern Operations.” Steam Info. Accessed February 2, 2021. https://steamdb.info/app/1076160/.
24. Atherton, Kelsey.
25. Atherton, Kelsey.
26. Brynen, Rex. “UK Fight Club.” PAX Sims, June 11, 2020. https://paxsims.wordpress.com/2020/06/11/uk-fight-club/.
27. “Data Science Virtual Machines.” Microsoft Azure. Accessed February 7, 2021. https://azure.microsoft.com/en-us/services/virtual-machines/data-science-virtual-machines/.
28. “2020 Department of Defense Artificial Intelligence Education Strategy.” The Joint Artificial Intelligence Center, September 2020. https://www.ai.mil/docs/2020_DoD_AI_Training_and_Education_Strategy_and_Infographic_10_27_20.pdf.
29. “2020 Department of Defense Artificial Intelligence Education Strategy.”
30. “2020 Department of Defense Artificial Intelligence Education Strategy.”
31. Ashoori, Maryam, Weisz, Justin.” “In AI We Trust? Factors that Influence Trustworthiness of AI-Infused Decision-Making Processes.” IBM. December 5, 2019. https://arxiv.org/pdf/1912.02675.pdf., 2.
32. “Advantage at Sea: Prevailing with All-Domain Naval Power.” Office of the Secretary of the Navy. December 2020. https://media.defense.gov/2020/Dec/16/2002553074/-1/-1/0/TRISERVICESTRATEGY.PDF., 22.
33. McKinney, Michael. “Warfighting First? Not so Much.” U.S. Naval Institute. May 2019. https://www.usni.org/magazines/proceedings/2019/may/warfighting-first-not-so-much
34. “Initial Report of the Dean’s Cyber Warfare Ad Hoc Committee.” The US Naval Academy. August 21, 2009. https://www.usna.edu/Users/cs/needham/CyberSecurityInitiative/USNACyberInitiativeInitialReport_USNA-CS-TR-2011-02.pdf#search=ns310., 76.
“Core Curriculum Review.” USNA Division of Seamanship and Navigation. March 2, 2005. https://www.usna.edu/Academics/_files/documents/sapr/ProDev_Core.pdf., slide 13.
35. “Wargaming at the Naval Academy.” Shipmate. The United States Naval Academy Alumni Foundation. February 2021., 25-26.
36. “Ready, Relevant Learning.” Naval Education and Training Command. Accessed March 19, https://www.netc.navy.mil/RRL/.

37. Gilday, 11.

Feature photo: A U.S. Naval Academy Midshipman conducts a simulated T-6B Texan II flight on a newly installed virtual reality trainer device at the U.S. Naval Academy during Aviation Selection Night at Dahlgren Hall. (U.S. Navy photo by Lt. Cmdr. Rick Healey/Released)

If You Build It, They Will Lose: Competing with China Requires New Information Warfare Tools

Naval Intelligence Topic Week

By Andrew P. Thompson

The Modern Fight

Written into the most recent National Security Strategy is the principle that Great Power competition will continue to play a major role in the shaping of our strategic priorities.1 As the Navy continues adapting to operations below the level of armed conflict, how we implement combat capability must adjust. China’s modernization of its Navy, enhanced with its desired use of Artificial Intelligence (AI), should catalyze change in our own development efforts. Its modernization initiative directly supports its system destruction warfare principle, which operationalizes a system of systems approach to combat. Confronting this style of warfare requires a new mindset, and the Information Warfare apparatus, of which Naval Intelligence is an integral part, must align itself appropriately to support this change. While the last century’s wars heavily favored attrition-centric warfare, 21st century Great Power competition requires the use of warfare that is decision-centric. The Information Warfare Community (IWC) support required for such an approach must capitalize on the use of new technologies, developed from industry, to aid commanders. Doing so will allow the IWC to provide decision-makers with the best advantages as fast as possible and the method to accomplish such a feat will determine both the IWC’s and Naval Intelligence’s legacy in this modern fight.

By the end of 2020, China is assessed to have 360 battle force ready ships compared to the U.S. Navy with 297.2 Projecting forward to 2025, China will have 400 battle force ships and 425 by 2030.3 In addition to the sheer size of its projected ship count, China is currently making strides to modernize its programs associated with anti-ship ballistic missiles, anti-ship cruise missiles, submarines, aircraft, unmanned aircraft, and command and control, communications, computers, intelligence, surveillance, and reconnaissance (C4ISR) tools.4 One supporting element in modernizing these programs is the Chinese utilization of AI. According to the Congressional Research Service, “the Chinese aim to use AI for exploiting large troves of intelligence, generating a common operating picture, and accelerating battlefield decision-making.”5 As opposed to the bureaucratic red tape that exists in much of the U.S. defense acquisitions process, few such barriers exist in China’s between its commercial, academic, military, and government entities. Therefore, the Chinese government can directly shape AI development to meet its desired need in whatever capacity it wants. To support this effort, the Chinese government founded a Military-Civil Fusion Development Commission in 2017 in order to rapidly transfer AI technology, from whatever source, directly to the military.6 In doing so, China is incrementally utilizing AI to enhance its conventional force modernization programs at a more rapid pace than one impeded by self-imposed bureaucracy.

AI Benefits/Issues

The advantages of AI technology apply no matter which nation develops it, allowing combat systems to react at gigahertz speed. With such a dramatic shift in the time scale of combat, the pace of combat itself accelerates.7 Additionally, military AI use can provide an augmentation option for long-term tasks that exceed human endurance. For example, intelligence gathering across vast areas for long durations becomes more manageable for human analysts when using AI.

In addition to the above advantages, AI directly confronts, and has the potential to make sense of, the tremendous amount of data for analysts to process. While the U.S. military operates over 11,000 drones, with each one recording “more than three NFL seasons worth” of high-definition footage each day, there are simply not enough people to adequately glean all possible actionable intelligence from such media.8 Similarly overwhelming are the 1.7 megabytes of information that the average human generates every second.9 Therefore, AI-powered intelligence systems may offer a way to sift through the resulting data repositories in order to better understand behavior patterns. Further, after a desired set of iterations, AI algorithms may feed further analysis that refines earlier conclusions, and ultimately provide an even better understanding of complex information for decision-making advantage.10 While promising, skepticism is necessary. Dr. Arati Prabhakar, a former DARPA Director, noted, “When we look at what’s happening with AI, we see something that is very powerful, but we also see a technology that is still quite fundamentally limited…the problem is that when it’s wrong, it’s wrong in ways that no human would ever be wrong.”11 Such skeptical risk, however, does not outweigh the possible benefits of AI’s development and use.

While the advantages of AI technology are clear, our adversary’s approach to how this development takes place merits discussion. The Chinese AI development framework can be corrupt and favor sub-par research institutions, resulting in potential overinvestment, producing unneeded and wasteful surpluses.12 Conversely, whatever advantage the U.S. retains in AI technology research due to China’s own domestic malfeasance can quickly diminish by way of industrial espionage. Despite agreeing to the U.S.-China Cyber Agreement, in which both sides agreed that “neither country’s government will conduct or knowingly support cyber-enabled theft of intellectual property,” it was reported to Congress that “from 2011-2018, more than 90 percent of the Justice Department’s cases alleging economic espionage by or to benefit a state involve China, and more than two-thirds of the Department’s theft of trade secrets cases have had a nexus to China.”13 Such actions, while not germane exclusively to AI development, clearly show an aggressive approach to technological progress with little regard for agreed-upon rules. When applied to AI research, such aggressiveness may result in less safe outcomes due to China’s tolerance for risk at the expense of speed. This may eventually result in the U.S. possessing more capable applications in the long-term.14 However, such optimism does not exempt the U.S. from adjusting to the modern concept of warfare for which China is rapidly developing AI in the first place.

System of Systems/System Destruction Warfare

The People’s Liberation Army (PLA) no longer sees war as a contest of annihilation between opposing forces. Rather, it sees war as a clash between opposing operational systems.15 Thus, China sees the victor in a war as the side who renders the other side’s systems ineffective, the ultimate goal of system destruction warfare. This model demands a joint force that utilizes numerous types of units from multiple services to continuously conduct operations across the battlefield.16 The past predicated that dominance in one or more physical domains was sufficient for warfighting success. As an example, 20th century thought suggested that air dominance was necessary to achieve land or sea dominance. Systems confrontation, on the other hand, predicates that warfare success requires dominance in all domains: land, sea, air, cyber, electromagnetic, and space.17 However, for such dominance to occur, the first domain necessitating control is the information one, as it is the nucleus that ensures everything else within the overall system correctly functions.18

To account for this dynamic force posturing in all domains, the PLA requires multidimensional and multifunctional operational systems. Such system permutations enable enough flexibility to adjust to newly developed technology.19 Correspondingly, a degree of malleability is built into the architecture of the PLA’s system categories of entities, structures, and elements. Entities include the weapon platform itself. Structures include the matrix style interlink that allows for coordinated functioning. Elements include the system’s command and control, protection, and maneuver capabilities. When intertwined, the resulting web of each system’s entities, structures, and elements provides redundancies that ensure the overall system is greater than the sum of its disparate parts.20 That said, each part is elastic enough that taking one part away from the web will not result in a total loss, while adding a part is equally non-destructive.

With these systems, the PLA seeks to strike four types of targets: 1) targets that interrupt the flow of information within an enemy’s system, such as key data links to a system’s command and control, 2) targets that degrade essential elements of an enemy’s system, such as a system’s firepower capability, 3) targets that interrupt the operational architecture of a system, such as the physical nodes of the essential elements (i.e. the firepower network), and 4) targets that interrupt the tempo of an enemy’s systems architecture, such as a system’s “reconnaissance-control-attack-evaluation” process that is inherent to all operational systems.21 Thus, the PLA seeks to operationalize its destructive warfare model by targeting what it perceives as the most vulnerable parts of its adversary’s infrastructure. By building flexibility into the design of units within its own system of systems (entities, structure, and elements) used to conduct this targeting, China’s system destruction warfare model accounts for loss while simultaneously adapting to new developments. Such an approach makes for a leaner, smarter, and dynamic force.

Decision-Centric Warfare/Our Response

In the current environment, Carrier Strike Groups are the Navy’s common force packages that deliver multi-mission units.22 These groups are vulnerable due to their size and aggregation, providing the perfect units for the PLA to target with its system destruction warfare model. Other services’ main force packages, such as the Army’s Brigade Combat Teams and the Marines’ Expeditionary Units, are also reflective of a vulnerable force borne out of the attrition-centric warfare model.23 While this legacy mindset worked in the 20th century, Great Power competition in the 21st century provides the requisite scenario to impose multiple dilemmas on an enemy to prevent it from achieving objectives. This decision-centric warfare approach, where making decisions faster than the adversary is paramount, is the cornerstone ingredient of the required methodology to confront China’s destructive warfare model.24 Having the Navy’s current force package, the Carrier Strike Group, utilize AI and autonomous systems is the means by which this new approach can be operationalized.

In addition to the benefits of AI discussed earlier, autonomous systems afford forces the ability to conduct more distributed operations by way of disaggregating capabilities of more traditional multi-mission platforms into a larger number of less flexible and less expensive systems.25 Use of these autonomous systems, on an as-available basis, is the hallmark standard of the decision-centric model. Thus, command and control of autonomous forces is based on communications availability, rather than a hardened command and control network. Decision-centric warfare assumes, and accounts for, contested and/or denied communications, as a commander will only possess control of forces that he/she actually can communicate with.26

From a decision-centric warfare model perspective, the current force’s Mission Command actually undermines its ability to make the necessary quickest decisions. It does so because the current command and control of forces is dependent on working communications, or extensively troubleshooting them, all of the time. To enable commanders to address this shortfall, the adoption of a new command and control structure that combines human command and AI-enabled machine control is necessary. Such a structure would combine a human’s flexibility and creativity with a machine’s speed and scale.27 Over time, as discussed earlier, human commanders could adjust machine recommendations, thereby forcing the machine to learn, increasing the commander’s confidence in subsequent recommendations when communications are limited.28 The net result of this feedback loop is a decision-making apparatus superior to an adversary’s. When applied to enemy systems attempting to target/destroy friendly force systems, the resulting quick decision-making effectively outmaneuvers the opposing side.

A key enabler of this quick decision-making rests with the advent of the Information Warfare Commander position on Carrier Strike Group staffs, which has gradually elevated the status of the Information Warfare Community (IWC) across the service. Along with this position, personnel within the Strike Group IWC Enterprise are key enablers who must recognize that their ability to leverage decision-making and combat capability lies with their ability to enable AI and autonomous systems of the future, combine this enabling with their own understanding of enemy intentions, and ultimately make recommendations to improve the commander’s decision cycle.

To achieve this, IWC personnel must be cognizant of new technologies on the rise within industry, where the most promising disruptive innovation trends reside that can meet these challenges. As the National Security Strategy states, “We must harness innovative technologies that are being developed outside of the traditional defense industrial base.”29 To this end, and to “harness innovative technologies,” an AI-industry sponsor must be assigned to each Carrier Strike Group Information Warfare Commander and his/her subordinate staff. This sponsorship program would enable IWC personnel the ability to incorporate the most modern AI technology into at-risk portions of their portfolios and define exacting requirements for new tools that are flexible enough for future progressive technological investment. While such innovation developments may surpass the tenure of the personnel assigned to the Strike Group staffs, the output of each team will aid future teams’ performance and eventually the Navy’s fighting ability. As such, after several iterations of afloat Strike Group staffs working with their respective industry sponsor, the result would be the promotion of tool production that aids the service in possessing the technological and decision-making edge…and ultimately play a direct role in future potential conflicts.

Getting to this point will require a new mindset for IWC personnel. Most do not possess acquisitions experience and most have not worked in positions that require technological innovation. To aid in not overburdening an IWC staff, the TYCOM should assign an Acquisitions Community sponsor to each Information Warfare Commander. This new combined team, comprised of the Strike Group IWC personnel, the AI-industry sponsor, and the TYCOM-approved Acquisitions Community sponsor, would seek to prototype tools/designs that attack key problem areas encountered by end users (i.e. the IWC personnel), as stated earlier. By swiftly deploying new operational concepts into potentially useable tools and products, the new decision-making infrastructure would support a warfare model fit to confront China’s today.

When compared to every other warfare area within the Navy, the IWC requires the most modern technological advances in the least amount of time. While other communities have proven processes and protocols in place to implement new technologies into their existing platforms, the IWC is simply too new and in too much need to benefit from these practices. This demands that the IWC business model be different, as Information Warfare Commanders need tools right now to effectively compete and win. Further, they must be the right tools where end users have a direct say in what they get.

Great Power Competition will dominate our military’s focus for the foreseeable future and the Information Warfare Community, including Naval Intelligence, must adjust accordingly. Understanding that China intends to enhance its military modernization efforts with AI, that it thinks differently about warfare in the 21st century, and that we need to modify our own warfare model to effectively respond, the Information Warfare Community’s newfound status should elevate new technologies into our Navy’s decision-making and combat DNA. The nation, and our Navy, cannot afford a misstep in this realm. The next major conflict will possess high stakes in the information domain where the Navy’s IWC will be at the forefront.

LCDR Andrew Thompson is currently serving at the USINDOPACOM JIOC. As a Surface Warfare Officer, he served aboard USS BOONE (FFG 28) as the Communications Officer, at Destroyer Squadron FIFTY as the Operations Officer, and at Naval Special Warfare Group ONE as the Middle East Desk Officer. As an Intelligence Officer, he has completed tours at the Office of Naval Intelligence, the Navy Cyber Warfare Development Group, and Carrier Strike Group TWELVE (as the Deputy N2). He holds a B.S. in Naval Architecture (USNA ’05), an M.S. in Mechanical Engineering (NPS), and an M.A. in National Security Studies (Naval War College). He holds subspecialties in African Studies and Space Systems, and has deployed to the SOUTHCOM, EUCOM, AFRICOM, and CENTCOM AORs. The views expressed in this article are his own, and do not reflect those of the Department of Defense or the Intelligence Community. 

Endnotes

1 Trump, Donald J., National Security Strategy of the United States of America, December, 2017, 27.

2 “China Naval Modernization: Implications for U.S. Navy Capabilities—Background and Issues for Congress.”

3 Ibid., 2.

4 Ibid., 3.

5 “Artificial Intelligence and National Security,” Congressional Research Service, November 21, 2019, 21.

6 Ibid., 21.

7 Ibid., 27.

8 Ibid., 28.

9 Ibid., 28.

10 Ibid., 28-29.

11 Ibid., 29.

12 Ibid., 23.

13 Ibid., 23.

14 Ibid., 23.

15 Engstrom, Jeffrey, How the Chinese People’s Liberation Army Seeks to Wage Modern Warfare, Santa Monica, CA: RAND Corporation, 2018, 10-11.

16 Ibid., 12.

17 Ibid., 13.

18 Ibid., 12.

19 Ibid., 13.

20 Ibid., 14.

21 Ibid., 16-18.

22 Clark, Bryan, Dan Patt, and Harrison Schramm. Mosaic Warfare: Exploiting Artificial Intelligence and Autonomous Systems to Implement Decision-Centric Operations. Center for Strategic and Budgetary Assessments, 2020, ii.

23 Ibid., iii.

24 Ibid., iii.

25 Ibid., v.

26 Ibid., v.

27 Ibid., vi.

28 Ibid., vi.

29 Trump, Donald J., National Security Strategy of the United States of America, December, 2017, 29.

Bibliography

“Artificial Intelligence and National Security.” Congressional Research Service. November 21, 2019. https://fas.org/sgp/crs/natsec/R45178.pdf

“China Naval Modernization: Implications for U.S. Navy Capabilities—Background and Issues for Congress.” Congressional Research Service. May 21, 2020. https://fas.org/sgp/crs/row/RL33153.pdf

Clark, Bryan, Dan Patt, and Harrison Schramm. Mosaic Warfare: Exploiting Artificial Intelligence and Autonomous Systems to Implement Decision-Centric Operations. Center for Strategic and Budgetary Assessments, 2020. https://csbaonline.org/uploads/documents/Mosaic_Warfare_Web.pdf

Engstrom, Jeffrey. How the Chinese People’s Liberation Army Seeks to Wage Modern Warfare. Santa Monica, CA: RAND Corporation, 2018. https://www.rand.org/pubs/research_reports/RR1708.html

Trump, Donald J. National Security Strategy of the United States of America. December, 2017. https://www.whitehouse.gov/wp-content/uploads/2017/12/NSS-Final-12-18-2017-0905.pdf

Featured Image: Sailors wearing gas masks operate a combat direction system console aboard the guided-missile frigate Handan (Hull 579) during a 4-day maritime training exercise conducted by a destroyer flotilla of the navy under the PLA Northern Theater Command in waters of the Yellow Sea from March 27 to 30, 2018. (eng.chinamil.com.cn/Photo by Zhang Hailong)