ABSTRACT

Visions of the future of military AI are evergreen, but the reality of military automation is more complicated. Information system performance is often more about the quality of people and organizations than the sophistication of technology. This is especially true of machine learning, which lowers the costs of prediction but increases the value of data and judgment. For commercial AI, economic institutions help to provide quality data and clear judgment. These enabling complements are likely to be missing or less effective in the contested environment of war. In other words, the economic conditions that enable AI performance are in tension with the political context of violent conflict. This strategic tension is likely to lead to several unintended consequences. These include unmanageable organizational complexity, as militaries and governments struggle to provide quality data and clear judgment, and strategic controversy, as adversaries target the data and judgment that become sources of strength for an AI-enabled organization. The irony is that increasing military automation will make the human dimension of war even more important.