Three Laws Blog: Law and Artificial Intelligence

Homayounnejad and Overill on Preventing Autonomous Weapon Systems From Being Used To Perpetrate Intentional Violations of the Laws of War @homayounnejad @KingsCollegeLon

Maziar Homayounnejad, King's College London, Dickson Poon School of Law, Department of War Studies, and Richard Overill, King's College London, Department of Informatics, have published Preventing Autonomous Weapon Systems from Being Used to Perpetrate Intentional Violations of the Laws of War as TLI Think! Paper 7/2018. Here is the abstract.

Autonomous Weapon Systems (AWS) are essentially weapon systems that, once activated, can select and engage targets without further human intervention. While these are neither currently fielded nor officially part of any nation’s defense strategy, there is ample evidence that States and defense contractors are currently developing AWS for future deployment. In the near-term, such weapons will mainly follow the wide-area loitering concept, which largely attacks targets from a distance. In the longer term, it is likely that militaries will develop and field bipedal / humanoid robots with a degree of manual dexterity, and the capability to undertake both rescue missions, and ‘close quarters combat’ with human subjects. This raises questions on how such machines may be used in times of armed conflict, occupation or internal strife. Given their potential capability for human-like physical interactions, there is a concern – following empirical research – that such machines may be deliberately programmed to commit inhumane acts against civilians, such as rape, broader sexual violence, and torture. Furthermore, as AWS units will be programmed to logically follow ex ante instructions, there is a real risk that such terrorizing acts will occur more easily, more frequently and, once initiated, indefinitely; potentially unconstrained by the moral or ethical brakes, or the physical limitations of human soldiers. However, there are also a number of legal and technical safeguards that may mitigate this risk, thereby realigning the technology of AWS with its original and intended purpose. These include: 1) capability limitations in the near-term; 2) positive design-led safeguards; 3) a ‘kill switch’ function; 4) pre-emptive legal safeguards to enhance non-proliferation efforts; and 5) digital evidence to aid the prosecution of war crimes. Importantly, none of these are individually watertight, but collectively they offer a package of safeguards that should sufficiently mitigate the risk of sexual violence, torture and other forms of repression using AWS.

Download the article from SSRN at the link.

Rabkin and Yoo on Striking Power: How Cyber, Robots, and Space Weapons Change the Rules For War @EncounterBooks

ICYMI:

Jeremy Rabkin and John Yoo, Striking Power: How Cyber, Robots, and Space Weapons Change the Rules for War (Encounter Books, 2017). Here from the publisher's website is a description of the book's contents.

Threats to international peace and security include the proliferation of weapons of mass destructions, rogue nations, and international terrorism. The United States must respond to these challenges by embracing new military technologies. Autonomous robots and cyber and space weapons can provide more precise, less destructive means to coerce opponents or end humanitarian disasters. Efforts to constrain arms innovation are not only unlikely to succeed in the long run, but may unduly inhibit responses to current challenges in the meantime. These new weapons can provide timely warnings to potential adversaries without causing death or personal injury on a large scale, thus promising less destructive conflicts overall.

 

 

 

 

Burri on International Law and Artificial Intelligence

Thomas Burri, University of St. Gallen, is publishing International Law and Artificial Intelligence in the German Yearbook of International Law (2017/2018). Here is the abstract.

This paper proposes five arguments about major aspects of artificial intelligence and their implications for international law. The aspects are: automation, personhood, weapons systems, control, and standardization.

Download the essay from SSRN at the link.

Joh on Private Security Robots, AI, and Deadly Force @elizabeth_joh

Elizabeth E. Joh, University of California, Davis, School of Law, is publishing 'A Certain Dangerous Engine': Private Security Robots, Artificial Intelligence, and Deadly Force in the University of California Davis Law Review (2017). Here is the abstract.

Robots equipped with artificial intelligence will transform existing notions of work in fields as diverse as fast food, health care, manufacturing and the military. A recent use of a remote controlled robot equipped with lethal force has raised the question of how police might use robots to supplement or replace existing police work. Those same questions apply equally to private individuals who will want security robots, some of which are already in development and for lease. What about a future in which people employ autonomous and lethally armed security robots for protecting their homes and themselves? How would courts characterize security robots? One possibility involves spring guns. While spring guns may not ultimately be well suited as a way to discuss security robots, it turns out that many of the questions considered in the spring gun cases may begin the framework that courts will have to develop as robots enter the security business.

Download the article from SSRN at the link.