The military and the computer sciences have always had close links to the extent that the early development of computers was virtually exclusively limited to military application. Alan Turing was one of the founders of modern computer science and the AI benchmark known as the 'Turing Test', while also being the war-time scientist who broke the German's Enigma code through the use of computers. Later, as computing power increased, as later predicted by Moore's Law, and accompanied by increased sophistication in program languages and development systems, the scope of military applications was to expand rapidly. For example, computers were soon being used to simulate nuclear escalations scenario and how the then nuclear arms race might be played out.
However, AI applications really started to come into focus when, in 1981, the Japanese announced a new generation of computers that would be capable of logic deduction. Although this project would ultimately fail, due to a lack of real understanding of the inherent complexity of AI, it did not stop the military funding of AI research. What actually changed was the the focus on more 'intelligent' computer systems that could affect the outcome of real-time battlefield decisions. This said, such pragmatism did not stop the pursuance of futuristic visions of battles being fought by autonomous hardware, inclusive of robots. Even so, a myriad of practical AI problems persisted, such as functional AI vision, which lead to problems not only associated with object avoidance, but the far more hazardous issues of friend/foe recognition. As a consequence, present-day AI developments may appear to be addressing much smaller, and possibly shorter terms goals, such as voice recognition systems and expert information systems. In this context, there has been considerable success in aiding the modern military pilot process and ever-increasing amount of real-time data received during operational missions. Equally, today, state-of-the-art military aircraft have become a highly complex integration of avionics, navigation, communications and weapon systems, which need to be controlled not only second-by-second, but millisecond-by-millisecond. As such, much of the real-time control of the aircrafts sub-systems has already been computerised. In addition, onboard information systems are starting to predict what the pilot should do in any given scenario, which one-day may renew the original vision of battles being fought by autonomous hardware. It is highlighted that this is a military vision, which many might wish never to come true.
However, irrespective of your possible feelings about the use of military technology, it is an area of applied science that cannot be ignored by virtue of its budget, which probably dwarves almost any other single source of AI research funding. When decisions have to be made very, very quickly, which involve the assessment of an enormous amount of information in a life and death situation; AI could provide the crucial difference to the military. From developing intricate flight plans to implementing complex supply systems or creating training simulation exercises to automated fighter planes, AI could be a key component of the military in the future. The scope of the global military budget is so large that the number of AI related projects are probably counted in the thousands; therefore, the selection of the following four areas of AI development constitutes a very small sample purely by way of illustration:
It is believed that the topics selected are representative of current and forward-looking ideas that reflect the evolution of military thinking towards AI. Given that the focus of this section is AI, and not current military thinking, the emphasis of the following sub-sections is very much on the general concepts and how they might contribute to the development of AI.