As the U.S. Department of Defense (DoD) seeks to increase funding for artificial intelligence (AI) technologies for defense and national security purposes, a new policy memorandum directs the DoD to take steps to ensure that AI is designed, developed, and deployed in a responsible manner.

In a May 26, 2021, memorandum titled “Implementing Responsible Artificial Intelligence in the Department of Defense,” Deputy Secretary of Defense Kathleen Hicks calls for the incorporation of responsible AI principles into the DoD’s AI requirements and acquisition processes. Ms. Hicks wrote: “As the DoD embraces [AI], it is imperative that we adopt responsible behavior, processes, and outcomes in a manner that reflects the Department’s commitment to its ethical principles, including the protection of privacy and civil liberties.”

The memorandum outlines six “foundational tenets” for the DoD to implement “Responsible AI” across the DoD. It also reaffirms the DoD’s AI Ethical Principles and confirms that they apply to all DoD AI capabilities of any scale, including AI-enabled autonomous systems. The memo directs the Joint Artificial Intelligence Center (JAIC) to coordinate the DoD’s strategy for Responsible AI.  

The DoD’s focus on the ethical and responsible development and use of AI has practical implications for companies developing AI solutions for the defense and national security market. Defense contractors should anticipate that responsible AI will be an important consideration in DoD procurements of AI and machine learning (ML) capabilities and R&D funding efforts.

This update provides an overview of the deputy secretary’s memorandum and its significance for defense contractors.

The Deputy Secretary Hicks Memorandum

In her memorandum, Deputy Secretary Hicks calls for a “holistic, integrated, and disciplined approach” for implementing responsible AI across the DoD. The memorandum establishes six foundational tenets by which the DoD will implement responsible AI:

• Ensuring a “disciplined” responsible AI governance structure and processes at the component- and DoD-wide levels “for oversight and accountability,” and clearly articulating DoD guidelines and policies on responsible AI to accelerate its adoption.
• Taking steps to ensure warfighter trust in responsible AI, including by providing education and training and establishing a test and evaluation and verification and validation framework that integrates real-time monitoring, algorithm confidence metrics, and user feedback.
• Developing tools and policies throughout the DoD acquisition lifecycle to synchronize the implementation of responsible AI for AI products.
• Incorporating responsible AI into all applicable AI requirements.
• Building a robust national and global responsible AI ecosystem to improve collaboration across government, the academy, and industry, and to “advance global norms grounded in shared values.”
• Building and retaining a responsible AI-ready workforce.

Under the memorandum, the JAIC director will develop, assess, and report on the implementation of responsible AI. The memorandum directs the JAIC to take four steps to accelerate the adoption and implementation of responsible AI across DoD, including establishing a Responsible AI Working Council that will provide recommendations on the integration of responsible AI into the acquisition requirements as well as supply chain risk management practices. A report on the Working Council’s recommendations is due by September 23, 2021.

The Working Council is required to develop a Responsible AI “Strategy and Implementation Pathway.” An Initial Pathway is due by August 24, 2021, and a Final Pathway is due by October 23, 2021.

Increased AI Investment and Responsible AI

Deputy Secretary Hicks’ memorandum comes at a time of increased defense investment in AI. DoD is seeking to leverage AI and other cutting-edge technologies such as hypersonics and quantum computing to develop and field warfighting advantages to U.S. military forces. AI is viewed as central to the U.S. strategic competition with China and Russia. There are signs of a growing federal market for AI. The Biden administration’s proposed $715 billion defense budget for FY 2022 includes $874 million in AI investment. According to DoD, its AI efforts now number more than 600, about 50% more than its FY 2021 efforts.

In its final report issued in March 2021, the National Security Commission on Artificial Intelligence—an independent panel established in the FY 2019 National Defense Authorization Act to review and to make recommendations to the President and Congress on AI for defense and national security applications—recommends that the U.S. government increase R&D funding for AI as part of an integrated national strategy to enable the United States to be “AI ready” by 2025. The Commission’s report also highlights the need for ethical and responsible use of AI and recommends, for example, that the United States work with allies to develop international standards of practice for the development, testing, and use of AI-enabled weapons systems.

The JAIC is seeking to implement DoD’s Ethical Principles into practice as it looks for ways to accelerate and scale the adoption of AI across DoD. The principles call for AI to be developed and deployed in a manner that is responsible, equitable, traceable, reliable, and governable. According to DoD, the principles are intended to promote innovative uses of AI while protecting civil liberties, privacy, and American values.

Over the past year, the JAIC has coordinated with other DoD components to implement the Ethics Principles and advance a holistic approach to responsible AI. The Ethical Principles have been included in certain AI contract requirements, including in recent JAIC procurements for AI test and evaluation and data readiness services. The JAIC also states that it aims to engage with commercial companies and other partners on ethics and responsible AI as part of its Other Transaction acquisition vehicle Tradewind which is focused on efficient AI acquisition processes.

Implications and Observations

The memorandum highlights the emerging importance of ethics and responsible AI for DoD, with important implications for how AI is acquired, developed, and used. Companies interested in selling AI technologies to DoD should anticipate that DoD’s Ethical Principles and responsible AI will be incorporated into agencies’ contract requirements and source selection evaluations and will result in new contract requirements for contractors. The JAIC and the Responsible AI Working Council will play key roles in DoD’s implementation process.

[View source.]