AI Bibliography

It is a time of rapid disruptive technological change, especially in the field of artificial intelligence (AI). While this technology has been developed by and for the commer- cial sector, the apparent potential for AI in military applications is now leading armed forces worldwide to experiment with embryonic, AI-enabled defence systems to determine how these could best be used for combat and peacetime tasks.

Australia is no different, with funding allocated in the 2020 Defence Strategic Update to begin introducing AI capabilities into Defence. This process will involve developing AI applications that address defined tactical-level and strategic-level military prob- lems, building a skilled AI workforce, engaging with partners and allies, integrating ethics into AI applications and conducting AI experiments. A tangible demonstration of this plan in action is the opening this year of the Defence Technology Acceleration ColLab at Fairbairn, Australian Capital Territory.

Peter Layton’s paper contributes to this broadly based movement by considering the role AI might play in future sea, land and air combat operations at the tactical and operational levels of war. This is a little examined area, as much of the discussion so far has focused on the key technological issues and concerns. These deliberations have indicated that AI might be a significant technology in future wars, but there remain numerous uncertainties. This paper provides a starting point from which to begin a debate that will help to resolve some of these uncertainties.

The paper argues that AI will infuse most military machines; however, its gen- eral-purpose nature means that it is likely to be employed initially within existing operational level constructs. Given this, AI’s principal warfighting usefulness for the short-to-medium term is in ‘find and fool’. AI, with its machine learning, is excellent at finding items hidden within a high-clutter background; in this function, it is better than humans and much faster. However, AI can be fooled through various means; its great finding capabilities lack robustness. These two key characteristics could have a dramatic effect when applied to current sea, land and air operational level thinking.