A top priority for the U.S. Department of Defense (DoD) is implementing responsible artificial intelligence (RAI). The DoD Deputy Secretary of Defense Kathleen Hicks released a new policy memorandum reaffirming the DoD AI Ethical Principles adopted in early 2020 for the design, development, deployment and use of AI capabilities.
The memorandum outlines the five principles for the ethical development of artificial intelligence capabilities:
- The first two principles — responsible and equitable — focus on ensuring that data sets are accurate and unintended bias is minimized. High-quality data and accurate data annotation are the foundations to develop successful AI applications.
- The next three principles — traceable, reliable, and governable — ensure that the workforce has an appropriate understanding of the technology, data, processes, and development.
In addition to the five ethical AI principles, the memorandum establishes six foundational tenets by which the DoD will implement responsible AI. The guidance requires disciplined governance, warfighter trust, a systems engineering and risk management approach to AI acquisitions throughout the product life cycle, the inclusion of RAI principles in appropriate DoD AI capabilities, a robust national and global RAI ecosystem, and an RAI-ready workforce.
Access the complete memorandum here: Implementing Responsible Artificial Intelligence in the Department of Defense.