Volume 17, No 1, Spring 2022 ISSN 1932-1066
LAWS and the Law: Rules as Impediment to Lethal Autonomy
Kevin Schieman
United States Military Academy, West Point
kevin.schieman@westpoint.edu
Abstract: The ability to follow the laws of war is arguably a necessary, if not sufficient condition for morally acceptable lethal autonomous weapons use. Yet following laws and other types of abstract rules is far more demanding than one may realize. This type of rule-following requires an agent to overcome three main difficulties: the heterogeneity of rules, semantic variation, and the demands of global reasoning. Taken together, these demands set an ambitious agenda for research focused on building machines capable of satisfying even a modest standard of legal compliance.
Keywords: Military ethics; machine ethics; lethal autonomous weapons; jus in bello; artificial intelligence; law; moral agency; machine agency.
language model with a somewhat general ability to complete a range of tasks in response to natural language prompts. It is hardly an exaggeration to say that the technology has caused a near panic in higher education and even led some to lament the impending death of the scholarly essay. Similarly important are advances in other domains of research such as self-driving cars, which appear to be making measurable progress in their capacity to cope with the irregularity and unpredictability of driving on public roads. It is in view of this progress that I want to revisit an important argument from the literature on Lethal Autonomous Weapon Systems (LAWS); a LAWS, according to the U.S. Department of Defense, is a system that
once activated, can select and engage targets without further intervention by an operator.2
2 Kathleen Hicks, Autonomy in Weapon Systems (DoD Directive 3000.09), Washington, DC: 25 January 2023, p. 21, https://www.esd.whs.mil/portals/54/documents/dd/issuances/dodd/300009p.pdf
Recent technological advances have led to a surge of optimism about the near-term prospects for many applications of artificial intelligence and machine learning.1 Especially significant have been breakthroughs in computer image recognition and natural language processing, hard problems with substantial practical utility that once seemed prohibitively difficult to resolve computationally. This progress has given rise to speculation that humans have reached an inflection point in artificial intelligence research where the technology has advanced to such a level that intelligent machines will begin to assume weightier roles in our lives. Perhaps nowhere has this been more obvious than with the public release of the artificial intelligence chatbot ChatGPT, a large
1 Disclaimer: The views expressed in this paper are those of the author and do not reflect the official policy or position of the United States Military Academy, the Department of the Army, the Department of Defense, or the United States Government.
Existenz
An International Journal in Philosophy, Religion, Politics, and the Arts
