The Pentagon is making a massive push to accelerate the application of artificial intelligence to ships, tanks, aircraft, drones, weapons and large networks as part of a sweeping strategy to more quickly harness and integrate the latest innovations.
Many forms of AI are already well-underway with US military combat systems, yet new technologies and applications are emerging so quickly that Deputy Secretary of Defense Patrick Shanahan has directed the immediate creation of a new Joint Artificial Intelligence Center.
“The Deputy Secretary of Defense directed the DoD Chief Information Officer to standup the Joint Artificial Intelligence Center in order to enable teams across DoD to swiftly deliver new AI-enabled capabilities and effectively experiment with new operating concepts in support of DoD's military missions and business functions." DoD spokeswoman Heather Babb told Warrior Maven.
Pentagon officials intend for the new effort to connect otherwise disparate AI developments across the services. The key concept, naturally, is to capitalize upon the newest and most efficient kinds of autonomy, automation and specific ways in which AI can develop for the long term -- yet also have an immediate impact upon current military operations.
AI performs a wide range of functions not purely restricted to conventional notions of IT or cyberspace; computer algorithms are increasingly able to almost instantaneously access vast pools of data, compare and organize information and perform automated procedural and analytical functions for human decision-makers in a role of command and control. While AI can of course massively expedite data consolidation, cloud migration and various kinds much-needed cybersecurity functions, it is increasingly being applied more broadly across weapons systems, large platforms and combat networks as well.
Rapid data-base access, organizing information and performing high-volume procedural functions are all decided advantages of AI applications. Algorithms, for example, are increasingly able to scan, view and organize ISR input such as images or video – to identify points of combat relevance of potential interest to a commander.
AI enabled technology can perform these kinds of procedural functions exponentially faster than humans can, massively shortening the crucial decision-making timeframe for combat decision makers. At the same time, many experts, developers and military leaders recognize that the certain problem-solving faculties and subjective determinations unique to human cognition - are still indispensable to decision making in war.
For this reason, advanced AI relies upon what developers refer to as “human-machine” interface or “easing the cognitive burden” wherein humans function in a command and control capacity while computer automation rapidly performs a range of key procedural functions.
AI & IT
This AI-driven phenomenon is of particular relevance when it comes to data systems, IT as a whole and advances in cybersecurity. For instance, Air Force developers are using advanced computer automation to replicate human behavior online – for the specific purpose of luring and tracking potential intruders. Also, AI can be used to perform real-time analytics on incoming traffic potentially containing malware, viruses or any kind of attempted intrusion. If the source, characteristics or discernable pattern of an attempted intrusion are identified quickly, cyber defenders are better positioned to respond.
When high-volume, redundant tasks are performed through computer automation, humans are freed up to expend energy pursuing a wider range of interpretive or conceptual work.
For example, the Army is working with a private firm called NCI to establish a certification of worthiness for a specific AI-enabled program designed to streamline a number of key tasks.
The NCI-developed program enables account creation, account deletion, background checks and other kind of high-volume data analysis.
“You can log into 10 different websites simultaneously, rather than having a person do that. A machine can go through and gather all the information for a person,” Brad Mascho Chief AI Officer, NCI, told Warrior Maven in an interview. “Humans can focus on higher priority threats.”
At the same time, big data analytics can quickly present new challenges for a variety of key reasons; a larger data flow can make it difficult for servers to “flex” as needed to accommodate rapid jumps in data coming through. Therefore, AI-empowered algorithms such as those engineered by NCI are needed to organize incoming data and identify anomalies or potential intrusions.
There is also a growing need for more real-time monitoring of activity on a message “bus,” because standard analytics methods based on probability and statistical probability often detect intrusions after the fact and are not always reliable or 100-percent accurate, cybersecurity experts and analysts explain.
AI & Navy Ships
The Navy is taking technical steps to expand AI and cyber harden its growing ship-based ocean combat network, called Consolidated Afloat Networks and Enterprise Services (CANES).
CANES is being installed on carriers, amphibious assault ships, destroyers and submarines, and the service has completed at least 50 CANES systems and has more in production, Navy developers said.
Upgraded CANES, which relies upon hardened cyber and IT connectivity along with radio and other communications technologies, is being specifically configured to increase automation - and perform more and more analytical functions without needing human intervention. It is one of many emerging technologies now being heavily fortified by new algorithms enabling artificial intelligence, senior Navy leaders told Warrior Maven.
Surface ships such as the Littoral Combat Ship, rely upon a host of interwoven technologies intended to share key data in real time - such as threat and targeting information, radar signal processing and fire control systems. CANES connectivity, and AI-informed analysis, can be fundamental to the operation of these systems, which often rely upon fast interpretation of sensor, targeting or ISR data to inform potentially lethal decisions.
The LCS, in particular, draws upon interconnected surface and anti-submarine “mission packages” engineered to use a host of ship systems in coordination with one another. These include ship-mounted guns and missiles along with helicopters, drones such as the Fire Scout and various sonar systems - the kinds of things potentially enhanced by AI analysis.
Navy developers say increasing cybersecurity, mission scope and overall resiliency on the CANES networks depends on using a common engineering approach with routers, satcom networks, servers and computing functions.
Increased computer automation also performs a large function on the Navy’s emerging Ford-Class aircraft carriers. The new carriers use advanced algorithms to perform diagnostics and other on-board maintenance and procedural tasks independently. This, Navy developers say, allows the service to reduce its crew size by as many as 900 sailors per carrier and save up to $4 billion dollars over the life of a ship.
AI & Cyber Defense
Algorithms calling upon advanced AI are being used to quickly access vast pools of data to perform real-time analytics designed to detect patterns and anomalies associated with malware.
“Every day, the Defense Department thwarts an estimated 36 million e-mails containing malware, viruses and phishing schemes from hackers, terrorists and foreign adversaries trying to gain unauthorized access to military systems,” Babb told Warrior Maven earlier this year.
One particular technique, now being developed by CISCO systems, seeks to address a particular irony or cybersecurity paradox; namely, while much DoD network traffic is encrypted for additional safety, encryption can also make it more difficult for cyber defenders to see hidden malware in the traffic.
CISCO is now prototyping new detection methods as part of an effort to introduce their technology to the US military services.
“We have the ability to read and detect malware in encrypted web traffic. Even though the data is encrypted there is still a pattern to malware,” Kelly Jones, Systems Engineer for CISCO Navy programs, told Warrior Maven.
AI & Large Combat Platforms, Tanks & Fighter Jets
Real-time analytics, informed by AI, has already had much success with both Army and Air Force Conditioned-Based Maintenance initiatives. The Army used IBMs Watson computer to perform real-time analytics on sensor information from Stryker vehicles and tactical trucks.
Drawing upon seemingly limitless databases of historical data, Watson was able to analyze information related to potential engine failures and other key vehicular systems. Properly identifying when a given combat-vehicle system might malfunction or need repairs helps both combat and logistical operations. Furthermore, the Army-IBM Stryker “proof of principle” exercise was able to wirelessly transmit sensor data, enabling AI to compare new information gathered against a historical database in seconds.
The Army is also working with IBM to test AI-enabled “autonomy kits” on tactical trucks designed to enable much greater degrees of autonomous navigation.
Advanced computer algorithms, enhanced in some instances through machine learning, enable systems such as Watson to instantly draw upon vast volumes of historical data as a way to expedite analysis of key mechanical indicators. Real-time analytics, drawing upon documented pools of established data through computer automation, can integrate otherwise disconnected sensors and other on-board vehicle systems.
“We identified some of the challenges in how you harmonize sensor data that is delivered from different solutions. Kevin Aven, partner and co-account lead, Army and Marine Corps, IBM Global Business Services, told Warrior Maven in an interview earlier this year.
Watson, for example, can take unstructured information from maintenance manuals, reports, safety materials, vehicle history information and other vehicle technologies – and use AI to analyze data and draw informed conclusions of great significance to military operators, Aven explained.
When created, IBM stated that, "more than 100 different techniques are used to analyze natural language, identify sources, find and generate hypotheses, find and score evidence, and merge and rank hypotheses," according to IBM Systems and Technology.
Working with a firm called C3IoT, the Air Force is doing something similar with F-16s. On board avionics and other technologies are monitored and analyzed using AI-enabled computers to discern when repairs or replacement parts are needed.
Applications of AI are also credited with enabling the F-35s “sensor fusion” technology which uses computer algorithms to autonomously gather and organize a wide-range of sensor data for the pilot.
It goes without saying that targeting data is of critical importance when it comes to mechanized ground warfare. With this in mind, Army combat vehicle developers are prototyping AI-enabled sensors intended to combine sensor information essential to identifying targets. If long-range EO/IR or thermal imaging sensors are able to both collect and organize combat data, vehicle crews can attack enemy targets much more quickly.
Some near-term applications, senior officials with the Army Research Laboratory say, include increased air and ground drone autonomy. It is an example of an area where AI is already having a large impact and is anticipated to figure prominently over the long-term as well.
“We know there is going to be unmanned systems for the future, and we want to look at unmanned systems and working with teams of manned systems. This involves AI-enabled machine learning in high priority areas we know are going to be long term as well as near term applications,” Karl Kappra, Chief of the Office of Strategy Management for the Army Research Lab, told Warrior Maven in an interview. “We also know we are going to be operating in complex environments, including electromagnetic and cyber areas.”
For instance, Kappra explained that sensor-equipped micro-autonomous drones could be programed with advanced algorithms to send back combat-relevant images or provide attacking forces with key interior dimensions to a target location.
“We are looking at micro-electrical mechanical systems and image-based systems to fly through a building autonomously and show you where walls and threats are inside the buildings,” Kappra said.
Also, Army combat vehicle developers consistently emphasize manned-unmanned teaming with “wing man” drone robots operating in tandem with manned vehicles to carry ammunition, test enemy defenses, identify targets and potentially fire weapons. Some senior Army weapons and technology developers have said that most future combat vehicles will be engineered with some level of autonomous ability or manned-unmanned teaming technology.
Warfare, Ethics & AI
Interestingly, debates about the future of AI, especially when it comes to autonomy, continues to spark significant controversy. Current Pentagon doctrine specifies that there must always be a “human-in-the-loop” when it comes to making decisions about the use of lethal force. However, the technology enabling an autonomous system to track, acquire and destroy a target by itself without needing human intervention – is already here.
In a previous interview with Warrior Maven, an Air Force scientist made the point that the current doctrine is of course related to offensive strikes of any kind, however there may be some instances where weapons are used autonomously in a purely defensive fashion. For instance, AI-enabled interceptors could be programmed to knock out incoming enemy missile attacks – without themselves destroying anything other than an approaching enemy weapon. In this instance, AI could serve an enormously valuable defensive function by performing intercepts exponentially faster than having a human decision maker involved.
Naturally, this kind of technology raises ethical questions, and some have made the point that even though the US military may intend to maintain a certain ethical stance – there is of course substantial concern that potential adversaries will not do the same.
Also, while often heralded as the “future” of warfare and technology, AI does have some limitations. For example, problems presented in combat, less-discernable nuances informing certain decisions, determining causation and the analysis of a range of different interwoven variables – are arguably things best performed by the human mind.
Many things in warfare, naturally, are often a complex byproduct of a range of more subjectively determined factors – impacted by concepts, personalities, individual psychology, historical nuances and larger sociological phenomena. This naturally raises the question as to how much even the most advanced computer programs could account for these and other somewhat less “tangible” factors.
