AI in War: Shall we play a game?

In a 1983 film called WarGames, a young hacker finds a back door into what he thought was a server with unreleased, computer games. Instead, it was a military super computer running what was supposed to be an AI agent whose purpose was to control real military operations. In the film, reality gets confused with game-playing and the “AI” reaches close to starting World War III. In the end, the AI system understands just in time that a global nuclear war has no actual winners and even makes a statement that “the only winning move is not to play”, stopping the launch.?

It has been 40 years since that film and today it seems it is more relevant than ever, as AI starts to sound more like reality than science fiction and quantum computing is also making steps forward (i.e. Many countries launching national quantum strategies). Although you don’t have to worry about AI starting a nuclear war soon (i.e. nuclear arsenal systems are offline systems for now), still artificial intelligence is increasingly being used in war and military operations, and could entirely change the nature of warfare soon. Currently, some of the application areas in the military are the following:

1.???Autonomous weapon systems (AWS): Divided into semi-autonomous (i.e. Human In The Loop), human supervised (i.e. Human On The Loop, N/A due to insufficient time for supervisor reactions) & fully autonomous (i.e. Human Out Of The Loop with no human intervention). AWS are considered by some as the 3rd revolution in warfare (i.e. after gunpowder & the atomic bomb). At the time of writing, no military is known to operate officially this last category (i.e. due to the lack of supporting tech) and while there are international efforts to heavily regulate all such category weapons, the great military powers are reluctant to ban them. For now AI is used to control drones (air-land-water) so as to collect intelligence, attack predefined targets autonomously & deliver supplies but always with a human in the loop:

• Assisting semi-autonomous vessels (e.g. U.S. destroyer Sea Hunter, Boeing’s Orca submarine).

• Assisting vehicles (e.g. Ripsaw M5 drone accompanying U.S. tanks & Russian Uranium-9 tank).

• Collaborative combat aircrafts controlled by manned aircrafts (i.e. a pilot from a Next-Gen fighter supported by an AI co-pilot) or from the ground (i.e. remotely piloted drones that have reduced the number of soldiers in combat by 90%) E.g. Kratos’ XQ-58A.

2.???Intelligence, Surveillance & Reconnaissance (ISR): AI can be used to analyse with lightning speed, large amounts of data from surveillance cameras, drones & other IoT devices to connect the dots, identify potential threats & make decisions on how to respond prior to the other side, offering a tactical advantage.?

Few examples are:

• Tiny reconnaissance drones with thermal cameras

• The U.S. Air Force semi-autonomous robotic dog from Ghost Robotics.

• Don’t forget also that companies like Boston Dynamics are making great progress in building exceptional robots that can approximate the mobility, dexterity and agility of people and animals.

In their most recent demo, the latest version of their robot (ie "Atlas"), is spinning, bouncing, jumping, pushing, throwing, and flipping. It is thus not strange that thoughts that come in mind are if such robots will at some point be combined with AI for military purposes & the potential dangers of that.?

• Ukraine is training neural networks with available footage to allow the drones to auto-detect soldiers & vehicles & speed up the OODA loop (Observe, Orient, Decide & Act).
• Special sensors were used as part of an acoustic-thermal artillery reconnaissance system that aimed to locate enemy artillery, mortars, anti-aircraft, or tactical-missile firing positions, without emitting radio waves in fully automatic mode.

3.?????Cybersecurity: AI is used to attack & protect military networks (i.e. ID patterns & malware to defend or use to them to attack) from cyber-attacks on critical infrastructure. In today's world, such attacks have a significant impact on military operations & national security:

• In Ukraine it is believed that many DDoS & other cyber-attacks, took place from Russia to steal information but also paralyze Ukraine’s communication network.
• On Jan 11, 2023 the US aviation system experienced an unidentified issue impacting the update of Notice to Air Missions (NOTAMs) & leading to the cancellation of flights across the country.
• According to Check Point, OpenAI’s ChatGPT is already being used in the modern cyber threat era to create (i.e. code generation) malware & similar tools for cyber-attacks by the ones who are less-skilled or the ones skilled but wanting to effortlessly produce multiple variations of threats. Cybercriminals from all over the world, including Russia are attempting to bypass restrictions that prevent them from accessing ChatGPT (i.e. incidents where hackers try to bypass IP, card, phone & other requirements).
• At the time of writing, there is also a claim by Chinese researchers to have broken the RSA public-key encryption system (i.e. based on properties of large prime numbers) by using a quantum computer. This system is used in many government & financial systems for secure data transmission & storage as well as to secure communications & protect sensitive information (i.e. credit card numbers, personal identification numbers, text messages, speech, etc.) from unauthorized access. If this proves to be true, it would be a critical cybersecurity threat affecting many areas including the military, all communications, databases, banks, financial markets, etc.?

Perhaps these examples also explain why besides USA, many other countries are rushing to improve in AI & Quantum computing technologies & why in late 2022, NATO tested AICA (i.e. Autonomous Intelligence Cyber Defence Agent), a prototype for defending against simulated cyber-attacks. NATO has also recently recognized that these & other emerging-and-disruptive technologies (EDTs) present both risks & opportunities. As a result, the Alliance is currently forming expert teams to adopt as well as to establish international principles of responsible use of these technologies & to maintain NATO’s technological edge.?

4.?????Logistics & supply chain management: AI can be used to optimize the movement of supplies and equipment (i.e. quickest & low risk routes) to ensure that troops have the resources they need by additionally predicting demand & thus boosting efficiency & effectiveness.

5.?????Medical treatment: AI can be used to analyse medical data to predict which patients are most at risk of complications, to help plan treatment, and to monitor the progress of recovering.

6.?????Predictive maintenance: AI can be used to predict when equipment is likely to fail, to schedule maintenance and repairs, and to optimize the use of resources.

7.?????Training and simulation: AI can be used to create realistic simulations of battlefield scenarios, to train troops, and to test new equipment.

In most of these categories, AI is already applied or being tested. Artificial intelligence is therefore already having an impact on military operations & these effects will increase in the coming years as technological advances (i.e. AI, Robotics, IoT, Quantum Computing, etc.) enable new weapon systems. While NATO and the U.S. support the development and deployment of some AI related weapons, this is not the case for the unsupervised versions of these systems, making sure that humans remain in the loop & that only responsible AI should be used. Nevertheless, the absence of legal guidelines regarding the use of AI in war & armed conflicts, has led to concerns about the increased militarization of AI:

• The potential for fatal mistakes that cause unintended harm.
• The possibility of AI-powered systems being hacked or manipulated with serious consequences. A dependence on this technology could leave military forces vulnerable if it is compromised or it malfunctions.
• An unofficial, new AI arms race between nations & the risk of war due to that. Similar to the arms race during the Cold War, nations might engage in competition to develop such AI systems.
• Ethical concerns such as the potential for machines to make decisions that violate human rights or the laws of war.