Algorithms of Armageddon: The Impact of Artificial Intelligence on Future Wars

256 pages

Chris O’Flaherty

Through a comprehensive dissection and explanation of current governmental uses of big data, machine learning, and artificial intelligence (AI), this book clearly sets out why algorithmic warfare is already upon us, the associated opportunities, and also many of the concurrent threats.

Noting that big data is the fuel of AI, especially ‘clean’ data (i.e. higher quality, audited and correctly labelled digital information), the early chapters review some key lexicographical terms necessary to understand the differences between machine learning (“the ability of computers to modify their decision making when exposed to increasing amounts of data”) and AI (self-learning by comparing predicted and actual outcomes, refined by self-adding programme lines of code through trial and to correct error).

As the depth of analysis duly increases, the authors note that development of AI in some of our adversaries is faster and more effective than here in the West, due to other nations being less fussed by legal intricacies or moral consensus. Whilst seen as necessary to keep in check the potential civilian downsides of algorithmic control, such legal and moral constraints are imposing artificial arms-development barriers on our own side. Only when the true scale of algorithmic warfare becomes obvious in a peer-on-peer conflict will the national security disadvantage of such control become evident – when it is too late to address it successfully. By way of a capstone quote, Galdorisi and Tangredi “assert that the United States is losing the AI arms race and must make an extraordinary effort to catch up.”

Addressing how best to exploit the opportunities of algorithmic war, a recurring theme is the utility of Human-Machine teaming, or “collaboration between warrior and weapon … that combines the critical social intelligence of a human with the processing speed of a machine.” This point is illustrated by the experiments in the years following the 1997 defeat of chess champion Garry Kasparov by the IBM supercomputer Deep Blue; super computers were subsequently beaten (comprehensively) when the outputs from a chess computer programme were interpreted by a knowledgeable human, who combined algorithmic information with intuition to optimise the conduct of the game against an opposing machine.

Their subsequent examination of AI in the context of the laws of war, including targeting rules such as those governing ‘necessity, distinction, proportionality, and precautions’, made for an interesting read, although some assertions in the text did not accord with a number of UK teachings or interpretation of International Humanitarian Law.

Of importance is their recognition that AI can be deceived, fooled, or just under-nourished to the point of poor decision making. The Clausewitzian fog of war, especially during periods of large-scale destruction of the digital sensors that feed AI, may duly take excess prominence in future campaigns, lest it leads to especially inaccurate (or even self-destructive) algorithmic decision making.

In the closing chapters the authors amplify their biggest concern – not that AI would control humans, but that some humans will use AI to control other humans. Examples already include the Chinese Communist Party’s use of AI surveillance and data processing to instantly generate ‘social credit scores’ for citizens, thus automatically determining rights to shop, travel or even to attend sporting events. The morphing of such social control AI into a viable military application, for example by an occupying army, is achievable with just a few lines of new code and an ‘issued’ mobile telephone for each citizen of a defeated population.

As a rather good sequel to the authors’ AI at War (reviewed in NR Vol 110 No1), this book provides a detailed Western view of the current state, and potential evolution, of military AI. It is thoroughly recommended to all those interested in the future conduct of war, and to those who will help evolve both the government policies and the military tactics needed to win.