Abhora Vacuum: the Status of Artificial Intelligence and Ai Drones Under International Law*
| Año de Publicación | 2013 |
| Páginas | 0014 |
Spring/Summer, 2013
John F. Weaver.
INTRODUCTION
On April 23, 2013, in response to the rapid development of autonomous weapons, Human Rights Watch (HRW) announced the launch of the Campaign to Stop Killer Robots.[1]The group, an international coalition of non-governmental organizations, is dedicated to convincing governments to ban such weapons.[2]These goals are consistent with HRW's report from December 2012, Losing Humanity: The Case Against Killer Robots, which outlines the group's "numerous legal, ethical, policy, and other concerns with fully autonomous weapons."[3] In particular, HRW points to international humanitarian law, which applies to situations of armed conflict and governs the conduct of hostilities and the protection of persons during conflict, [4] as prohibiting "killer robots."[5]
HRW describes killer robots as weapons that are "able to select and engage targets without human intervention."[6] Not surprisingly, "killer robots" is not a phrase embraced by everyone interested in this technology. Neutral observers are more likely to use "autonomous drones." I prefer the term "AI drones, " because what we are talking about is technology that makes decisions, replicating an aspect of human intelligence - for better or worse - which is exactly what artificial intelligence (AI) does.
HRW's public campaign against AI drones coincides with the American government's public re-examination of its drone practices. On May 23, 2013, President Obama delivered a speech at the National Defense University that discussed American drone policy. Although noting the "precision of drone strikes, " he warned that drone strikes are not a "cure-all for terrorism" and that a drone-based "perpetual war... will prove self-defeating, and alter our country in troubling ways."[7] Similarly, in November of 2012, the U.S. Department of Defense issued Directive 3000.09, which requires that autonomous weapon systems "shall be designed to allow commanders to exercise appropriate human judgment over the use of force" and shall" [co]mplete engagements in a timeframe consistent with commander and operator intentions and, if unable to do so, terminate engagements or seek additional human operator input before continuing the engagement."[8]
Note that the Department of Defense and Human Rights Watch have reached two very different conclusions about AI drones under international law. HRW argues in Losing Humanity that AI drones violate international law. The Department of Defense, through Directive 3000.09, relies on the assumption that international law permits AI drones if they're overseen, regulated, and tested appropriately. This discrepancy is possible because international law does not actually directly address artificial intelligence, either with regard to drones or any other type of machine, vehicle, technology, etc. International law, like almost all forms of law, assumes that human beings make decisions. It does not address situations in which decisions are made by machines without human input.
This article uses AI drones to briefly explore that empty space. The very structure of international law - how it is created, how it is enforced, its historical development - means we are unlikely to have clear international law governing artificial intelligence any time soon. This represents a serious absence as numerous products and machines are set to rely on AI, including recent newsworthy examples like autonomous cars.
I. BRIEF HISTORY OF INTERNATIONAL LAW AND WEAPONS
Early International Law to Today's Treaties and Customs
International law has a deep history. Almost as soon as civilizations developed, they became interested in relations with the outside world.[9] Early religious and secular writings reference peace treaties and alliances between different sovereign peoples: Jews, Romans, Syrians, Spartans, Carthaginians, German tribes, Arab tribes, etc.[10] The Roman Empire conducted extensive treaties with its neighbors, although the Romans frequently did this as a first step toward absorbing the other p arty into the Empire, or as formal recognition of the other party as a vassal to the Empire.[11]
Treaties recovered from early civilizations in Asia, Africa, and Europe show that the societies on those continents were concerned with many of the same topics simultaneously, but without interaction among many of the peoples. Ambassadors, extradition of criminals, protection of foreigners, and international contracts were all issues that early societies addressed through treaties. The principles that came to govern these matters are remarkably similar, regardless of whether the civilizations were in the Mediterranean region or on the Pacific Ocean.[12]
The Romans developed a concept that attempted to encapsulate these universal principles: jus gentium, the body of law that is common to all people. Long after the Empire disintegrated and the concept of nations and states developed, jus gentium became one of the primary principles of international law, referring to universal laws and customs that have been consented to by all peoples and states throughout the world.[13] The combination of jus gentium and the patchwork of treaties between empires, tribes, and peoples constituted the general extent of international law until the mid-17th century.[14]
Today, the application of customs and treaties is still the primary force governing international law, although general principles of law recognized by nations and opinions of respected scholars are also relevant.[15] However, this system of law, by its nature, lacks a government that has the absolute right to pass and enforce laws affecting nations. Rather, international law depends on the nations themselves to enter into the process of making international laws willingly and then acting in good faith to enforce those laws within their borders and among themselves. In other words, each nation has to consent to being governed by international law, even when the potential outcome isn't what that nation wants.
Despite the apparent limitations, the international community has used the main elements of international law - treaties and customs - to govern controversial weapons. Those measures are useful when considering how international law might be used to govern AI drones.
Treaties Governing Weapons
The United Nations uses a variety of tools to encourage actions from its member states, including "norm-creating" treaties and resolutions from the Security Council. These forms of UN action permit it to respond quickly to technological developments.[16] For example, Resolution 1 of the United Nations General Assembly in 1946 responded to concerns related to atomic weapons, which were first used the year before. The resolution created the United Nations Atomic Energy Commission to address problems connected to atomic energy and permit its use only for peaceful purposes.[17] Subsequently, many nations signed multilateral treaties regulating and banning nuclear weapons, including the Treaty on the Non-Proliferation of Nuclear Weapons in 1970[18] and the Treaty Banning Nuclear Weapon Tests in the Atmosphere, in Outer Space and Under Water in 1963.[19]
More recently, the Convention on the Prohibition of the Use, Stockpiling, Production and Transfer of Anti-Personnel Mines and on their Destruction (the Landmine Convention) became effective in 1999, [20] in response to the number of injuries and deaths caused by landmines, particularly in unstable areas of the world.[21] The convention, by its terms, relies on the UN for assistance in worldwide acceptance and compliance.[22]
The treaties and conventions addressing nuclear weapons and landmines are significant because they represent deliberate actions by the international community to both establish law governing new weapons (nuclear arms) and change the existing customary international use of weapon technologies. Similar treaties may be necessary for AI drones (landmines).
International Customs Governing Weapons
Historically, custom governed much of international law, including conduct on the ocean, use of airspace and outer space, diplomatic immunity, and the rules of war.[23] Although there are now tens of thousands of treaties and international agreements, there is still room for custom to dictate international law.
First, custom guides treaty interpretation. Second, treaties never bind all nations.[24] Take the Landmine Convention, for example. The United States is still not a party to the Landmine Convention.[25] Although the U.S. has not used landmines since 1991 [26]it is free to discontinue that policy and resume following customary use of landmines. This includes using landmines to reduce the mobility of enemy personnel, channeling them into specific areas or scattering them over a broad area. Nations use landmines to disrupt the formations and delay the movements of hostile forces. Additionally, mine fields have been used to protect borders as a cost-effective solution to shortages of soldiers.[27]
However, the way customary international law permits the U.S. to use landmines is the very reason many people and nations believed the Landmine Convention banning landmines was necessary.[28] During the last 20 years of the 20th century, landmines claimed 1 million victims.[29] Somewhere between 75 and 85 percent of the victims were civilians.[30] Many of those were children. In some mine-infested areas, people are so desensitized to the danger that kids use the mines as wheels for toy trucks. Additionally, the countries most likely to be negatively affected are developing nations that depend on agriculture for survival, where mines can destroy huge tracts of productive soil.[31]
Conversely, questions about the customary use of...
Para continuar leyendo
Solicita tu prueba