Moderator: Professor Claire Finkelstein
Discussion Leader: Principal Deputy Assistant Secretary Paul Dean

Artificial intelligence (AI) is a transformative technology that has profound implications for national security and defense. In this session, participants will reflect on AI as a tool which can enhance the capabilities and performance of military and intelligence systems. This will touch on intelligence gathering and analysis, military decision making, the use of semi-autonomous and autonomous vehicles, weapons systems, and command and control. Participants will also discuss how AI poses significant challenges and risks, such as the management of bias, other ethical, legal, and moral dilemmas, technical vulnerabilities, and global competition, as well as the balance between commercial and government funding for the development of AI. How can the United States and its allies harness the benefits of AI while mitigating its dangers? What are the best practices and policies for developing and deploying AI in a responsible and effective manner? How can the United States and its allies cooperate and coordinate on AI issues to ensure interoperability, trust, and the alignment of interests?

Moderator: Dean Kevin Govern
Discussion Leader: Director Jonathan Elliott

Autonomous Weapon Systems (AWS) are machines that can select and engage targets without human intervention, using AI and machine learning algorithms. Participants in this session will discuss the potential of AWS to enhance military capabilities, efficiency, and performance in armed conflict, as well as to reduce human casualties and errors in fighting “cleaner” wars. This session will explore how AWS, and nuclear weapons more narrowly, pose significant legal, ethical, and humanitarian challenges. These include ensuring compliance with the law of armed conflict (LOAC), maintaining human control and responsibility, preventing escalation and instability, and addressing moral and social concerns. How can states and other actors develop and use AWS in a way that respects LOAC, acknowledges the varying capabilities between states, and protects civilians in armed conflict? What are the best practices and policies for ensuring accountability, transparency, and the reliability of AWS? How can states and other actors cooperate and coordinate on AWS issues to prevent arms races, foster trust, and promote peace and security? With respect to nuclear weapons, how can the international community ensure that AI is used in a responsible and effective manner to enhance nuclear governance and stability? What are the best practices and policies for developing and deploying AI in compliance with the existing nuclear norms and regimes—including those codified in President Biden’s Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence, the DoD’s Ethical Principles for Artificial Intelligence and Responsible Artificial Intelligence Strategy and Implementation Highway, and the State Department’s Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy—and in adherence with such conventions as the principle of distinction under international law? How can states forestall the extreme risk of the conversion of their nuclear arsenals into AI governed systems?

Moderator: Professor Michael Meier
Discussion Leader: Mr. Dan Stigall

The Department of Defense (DoD) has recently released its Civilian Harm Mitigation and Response Action Plan (CHMR-AP), which lays out a series of major actions DoD will implement to mitigate and respond to civilian harm resulting from U.S. Military operations. The plan recognizes that protecting civilians from harm is not only a moral imperative, but also a strategic necessity, as civilian harm can undermine operational effectiveness, create grievances, and erode legitimacy. One of the key elements of the CHMR-AP is to leverage emerging technologies such as artificial intelligence (AI) to enhance DoD’s ability to prevent, assess, and respond to civilian harm. This session will explore how AI can help DoD achieve its CHMR objectives, and will address the following questions: How can AI improve DoD’s understanding of the civilian environment and the potential sources and pathways of civilian harm? How can AI enhance DoD’s decision-making and targeting processes to minimize the risk of civilian harm? How can AI assist DoD in conducting timely and thorough assessments and investigations of civilian harm incidents? How can AI facilitate DoD’s response and remediation efforts when civilian harm occurs, such as providing ex gratia payments, condolence messages, or apologies? What are the best practices and policies for developing and deploying AI for civilian protection in a responsible and effective manner? What are the ethical, legal, and moral challenges and dilemmas of using AI for civilian protection? How can DoD cooperate and coordinate with allies, partners, and civil society on AI and civilian protection issues?

Moderator: Dr. Jean-Luc Cambier
Discussion Leader: Dr. Daniel Koditschek

The embodiment of increasingly autonomous capabilities in small, advanced mobile robots promises to deliver huge benefits. These same prospects also raise concerns about the possibility of their causing unethical or illegal harm along the spectrum of their use cases ranging from the well-discussed, but still perplexing, challenges associated with military applications through their potential deployment by police forces and extending to their eventual ubiquitous penetration into civilian spaces. Some companies at the forefront of these technologies have begun to develop a range of official policies in response to the novel mix of clear benefits and potential harms they promote. For example, Boston Dynamics’ public Statement of Ethical Principles explicitly eschews weaponization of their products. A group of similarly prominent small, advanced, mobile robot manufacturers (such as Agility Robotics, Anybotics, and Unitree) have joined Boston Dynamics in signing an open letter pledging not to weaponize their hardware or software products. This letter acknowledges the legitimacy of existing technologies applied in the interests of upholding laws and in national defense. Indeed, the U.S. Department of Defense has been a huge investor in—and is arguably largely responsible for—the development of many of these new as well as existing technologies. The specific combination of capabilities that now trigger recent concerns can be difficult to identify, and the responsibilities of commercial developers to address them remain very unclear.  What, if anything in today’s advanced platforms, is so different from robots that have been developed for and deployed in defense applications for decades? What are the best practices for the advanced mobile robotics industry in relation to this emerging spectrum of opportunities and challenges? What assistance and collaboration from robotics researchers and experts in policy and ethics would be helpful in better understanding the nature of possible harms, as well as the obligation and opportunities for industry to mitigate them?

Moderator: Professor Peter Asaro
Discussion Leader: Dr. Lisa Titus

Given the mix of great benefits and potential harms from the rapidly improving capabilities of advanced mobile robots, how can designers and developers ensure that their work is aligned with the greatest human good? A great majority of U.S. (and many international) robotics researchers enjoy substantial support from the military. It seems fair to assert that many of the impressive technologies we now see emerging from academic and corporate laboratories would not have been possible without that support. Scientists and engineers have had to balance their need for research funding with their responsibility to promote social good for many decades. What are the similarities or differences, for example, between the situation of physicists at the dawn of the atomic age and today’s AI and robotics researchers whose fundamental research seems to have produced technologies that may be crossing some qualitatively new threshold of capabilities? What are these capabilities, and how can their creators ensure they line up with human needs and goals? What are the opportunities for collaborating with industry and policy makers to better define and implement this alignment?  What responsibilities and possible obligations are incumbent upon researchers, teachers, and their students for helping develop such principles and putting them into practice? How can ethical thinkers and legal scholars aid in addressing these questions?

Moderator: Ms. Jessica Rajkowski
Discussion Leader: Professor Finkelstein

How do well-established international frameworks, such as LOAC and IHL, impact the work of robot designers and developers whose technologies may end up in military applications? What, if anything, is new about the nature of their fundamental research and corporate products that might require further development of such frameworks? How can the insights and expertise of roboticists be most usefully recruited to help advance progress in conceptualizing the new frameworks that may be needed. How can roboticists support international negotiations aimed at ensuring that militarized robots fit within the bounds of existing or suitably revised international frameworks? What principles exist—and what new ethical principles or laws may be required—to articulate guidelines and obligations bearing on the design and sale of robots to mitigate potential harm upon their entry into the civilian sphere? How can roboticists contribute usefully to the development and application of such principles? Given the increasingly ubiquitous nature of these machines, what best practices should be immediately adopted by researchers and developers to align robot capabilities and behaviors with human needs and goals in civilian spaces?