AI integration into military systems is inherently destabilizing.
James Johnson, 2020 (Postdoctoral Research Fellow at the James Martin Center for Nonprolifera¬tion Studies, WASHINGTON QUARTERLY, “Artificial Intelligence in Nuclear Warfare: A Perfect Storm of Instability?”)
Non-human decision-making As AI systems become more integrated into the strategic decision-making process, the potential vulnerabilities and risks associated with military autonomy will likely increase. According to former DARPA director Arati Prabhakar, despite AI’s potential tactical and operational impact, the technology remains “fundamentally limited” due to its propensity to misidentify objects and be easily spoofed.19 Prab¬hakar adds that decision-making errors committed by AI systems are often inex¬plicable and unlike errors humans would make. In one example, Prabhakar demonstrated how a picture of a baby holding a toothbrush was misidentified by a machine-learning algorithm as a baby with a baseball bat.20 Unpredictable errors and decisions made by AI systems deployed at scale and across multiple domains and combat theaters would be inherently destabilizing.21