Artificial Intelligence for Military Decision Support

Authors:

  • Johan Schubert

Publish date: 2018-02-26

Report number: FOI-R--4552--SE

Pages: 48

Written in: Swedish

Keywords:

  • Artificial Intelligence
  • Machine Learning
  • Neural Networks
  • Deep Learning
  • eXplainable AI
  • XAI
  • Military Decision Support
  • Military Command and Control.

Abstract

In this report we provide an overview of methods and applications with artificial intelligence for military decision support. The focus is on the tactical and operational levels. Applications are available in analyzing large amounts of surveillance data, intelligence data, etc., and support for planning and execution of operations at tactical and operational levels. The analysis is based on open publications in the scientific literature in the field of artificial intelligence for military decision support and associated areas of machine learning and deep learning. The analysis is grouped into three subdivisions: (i) Descriptive decision support, (ii) Predictive decision support, and (iii) Prescriptive decision support. These three areas answer questions of the type: (i) What is the situation, (ii) What will happen, and (iii) What action options do I have and how do I act? The study is based on Horizon Scanning that include at least one of the scientific terms artificial intelligence, machine learning or deep learning combined with at least one of the military terms of military, defense and command and control. The searches have been done using FOI's computer tool HSTOOL. A conclusion from the study is that artificial intelligence today is strong on all fronts. Artificial intelligence includes many different methodologies; automated reasoning to draw conclusions and provide recommendations, approximate reasoning dealing with uncertain information, and machine learning to let computers learn from examples or from feed-back from computerized games or simulation systems. The hottest area within artificial intelligence is machine learning, and especially deep learning. Using machine learning, however, will not be sufficient as a means of military decision support. It is necessary to take the results of machine learning and include them in the decision support systems that interact with information from automated and approximate reasoning and information fusion. The aggregate results are compiled and presented to the decision makers. What we should expect in the next decade is a very large number of successful applications of AI in many different defined areas of application in both the civilian and military sectors. For the military sector, the benefit of artificial intelligence is to deliver decisive support when time is too short or when the number of choices is too big for people to analyze all alternative actions.