REAL RISKS: UNDERSTANDING—AND MITIGATING— VULNERABILITIES IN THE MILITARY USE OF AI

Abstract

No one likes to wake up in the morning, but now that artificial intelligence–powered algorithms set our alarms, manage the temperature settings in our homes, and select playlists to match our moods, snooze buttons are used less and less. AI safety-assist systems make our vehicles safer and AI algorithms optimize police patrols to make the neighborhoods we drive through, and live in, safer as well. All around us, AI is there, powering the tools and devices that shape our environment, augment and assist us in our daily routines, and nudge us to make choices about what to eat, wear, and purchase— with and without our consent. However, AI is also there when our smart devices start deciding who among us is suspicious, when a marginalized community is disproportionately targeted for police patrols, and when a self-driving car kills a jaywalker.