Military inventions often find their way into civilian life. The opposite is also true, for example AI-based drones that enable aerial photography for agriculture. When these systems are modified and used militarily, they are so-called dual technologies. The same applies to facial recognition systems that are integrated into various weapon systems. There are no international regulatory mechanisms for AI-based weapon systems. Even if there were, such treaties would lag behind reality. Because AI has already arrived in modern armies. Germany, Great Britain and Italy are already developing LAWs, write authors Justin Haner and Denise Garcia in the scientific journal Global Policy. The Chinese military is also testing so-called swarm technologies. These make it possible to coordinate and maneuver over a thousand drones using AI – without any human intervention. Ironically, this technology was originally developed to fight fires in cities and is therefore also a dual technology.
With every step towards complete autonomy of these weapons systems, the possibilities for human intervention decrease. While AI-supported pilot assistance systems initially only warned soldiers of potential dangers, the possibilities are now more extensive: The AI can not only recognize the danger, but also act in place of the pilot – provided no human intervenes, according to a scientific article by the Peace Research Institute Oslo. But what if the AI system mistakes a civilian vehicle for a personnel carrier? How long will it take until the intermediate step of human intervention is skipped in favor of faster reaction? Of course, a human can also mistake a military vehicle for a civilian truck. However, with a shift in competence to AI, another actor is created who, in addition to the expected human errors, can represent another potential source of error.
AI-supported military technologies have long been part of modern armed forces. It is unlikely that a state will give up these capabilities, especially if potential military opponents do not. The digitalization of the modern battlefield requires adjustments. Military operations are being carried out more quickly and precisely, so that a person is no longer able to select individual targets when confronted with a swarm of drones, for example, concludes a position paper from the Fraunhofer Defense, Prevention and Security division. The use of AI systems forces military actors to counter these systems with their own AI technology. The result: a loss of competence in favor of AI.
If there are no international non-proliferation treaties in the near future, further AI systems will have to be developed to be able to respond to the technologies currently available. The spread of AI by state and non-state actors cannot be prevented by regulation. However, this realization may come too late. Houthi rebels are already using armed drones and the terrorist group Boko Haram has also used drones with improvised explosive devices. The next step towards the use of AI-controlled drone systems by non-state groups is not as far away as it was a few years ago. From that point on, at the latest, it would be unacceptable for national armed forces to fall behind in these matters.
Learn more
A digital training course on the topic of “AI in the military” took place in mid-July – read about it here.
