The plan establishes a norm for military AI – it sends a strong signal of the importance of safety and reliability – following through with actions is crucial to uphold our leadership
Horowitz and Scharre 2021 - Director of the Emerging Capabilities Policy Office in the Office of the Under Secretary of Defense for Policy and Vice President and Director of Studies at CNAS [Michael, Paul, January 12 2021, “AI and International Stability: Risks and Confidence- Building Measures”, https://www.cnas.org/publications/reports/ai-and-international-stability-risks-and-confidence-building-measures, Acc 6/18/22. M.A.]
Promoting Norms In 2019, the U.S. Defense Innovation Board proposed a set of AI principles for the U.S. Defense Department, which DoD subsequently adopted in early 2020. While these principles no doubt have domestic audiences in the U.S. defense community and tech sector, they also serve as an early example of a state promulgating norms about appropriate use of AI in military applications. The DoD AI principles included a requirement that DoD AI systems be responsible, equitable, traceable, reliable, and governable. (The full set of DoD AI principles is included in the Appendix). Similarly, the DoD’s unclassified summary of its AI strategy, released in 2019, called for building AI systems that were “resilient, robust, reliable, and secure.” A focus of the strategy was “leading in military ethics and AI safety.” There is value in states promoting norms for responsible use of AI, including adopting and employing technology in a way that reflects an understanding of the technical risks associated with AI systems. While stating such principles is not the same as putting in place effective bureaucratic processes to ensure their compliance, there is nevertheless value in states publicly signaling to others (and to their own bureaucracies) the importance of using AI responsibly in military applications. While these norms are at a high level, they nevertheless signal some degree of attention by senior military and civilian defense officials to some of the risks of AI systems, including issues surrounding safety, security, responsibility, and controllability. These signals may aid internal bureaucratic efforts to mitigate various AI-related risks, as bureaucratic actors can point to these official documents for support. Additionally, to the extent that other nations find these statements credible, they may help signal to other nations at least some degree of awareness and attention to these risks, helping to incentivize others to do the same. One risk to such statements is that if they appear manifestly at odds with a state’s actions, they can ring hollow, undermine a state’s credibility, or undermine the norm itself. For example, loudly proclaiming the importance of AI ethics while using AI systems in a clearly unethical manner, such as for internal repression or without regard for civilian casualties, could not only undermine a state’s credibility but also undermine the value of the norm overall, especially if other states fail to highlight the disconnect. Following through with meaningful actions to show how a state puts these norms into practice is essential for them to have real value.