Autonomous Weapons and International Humanitarian Law: Advantages, Open Technical Questions and Legal Issues to be Clarified

• Author: Marco Sassoli
• Position: Professor of International Law at the University of Geneva, Switzerland, Director of the Department of Public International Law and International Organization, Associate Professor at the Universite du Quebec, Montreal, Canada and Commissioner within the International Commission of Jurists
• Pages: 308-340

International Law Studies 2014

308

A

I

Autonomous Weapons and International

Humanitarian Law: Advantages, Open Technical

Questions and Legal Issues to be Clarified

Marco Sassóli*

I. INTRODUCTION

utonomous weapons are subject to great controversy. They have been

defined as weapon systems

1 “that can learn or adapt [their] functioning in

response to changing circumstances in the environment in which [they are]

deployed.”2 Such systems, once developed, should, through sensors that

give them situational awareness, be able to identify both legitimate targets

and hopefully civilians/civilian objects that may potentially suffer incidental

* Professor of International Law at the University of Genev a, Switzerland, Director

of the Department of Public International Law and International Organization, Associate

Professor at the Universite du Quebec, Montreal, Canada and Commissioner within the

International Commission of Jurists. The author would like to t hank Ms. Yvette Issar,

doctoral candidate at the University of Geneva, for her valuable research assistance, criti-

cal remarks and for having revised this text.

1. Hin-Yan Liu, Categorization and Legality of Autonomous and Remote Weapons Systems, 94

INTERNATIONAL REVIEW OF THE RED CROSS 627, 635–36 (2012).

2. International Committee of the Red Cr oss, International Humanitarian Law and the

Challenges of Contemporary Armed Conflict: Report Prepared for the 31 st International Conference of the

Red Cross and Red Cre scent 39 (2011), available at http://www.icrc.org/eng/resources/docu

ments/report/31-international-conference-ihl-challenges-report-2011-10-31.htm.

Autonomous Weapons and IHL: Advantages, Questions and Issues Vol. 90

309

effects of attack. Identification would then trigger corresponding action

through processors or artificial intelligence that would “decide . . . how to

respond . . . and effectors that carry out those ‘decisions.’”

3 Ideally, auton-

omous weapons would select and engage targets without ongoing human

intervention in an open environment under circumstances which are un-

structured and dynamic. At present, no weapon system possesses such ca-

pabilities. The absence or presence of human intervention is, however, a

relative distinction, as is the distinction between humans “in,” “on” or “out

of the loop,” which is therefore not very helpful.4 Despite the system’s au-

tonomy, human beings will inevitably be involved, either in overseeing the

operation of the weapon, or at least in producing and programming the

weapon systems. There is agreement that, although these systems do not

yet exist, they could be developed within twenty years. Many request that

they be purely, simply and preventively banned, specifically because their

use would not be consistent with international humanitarian law (IHL).5

The 117 States parties to the UN Conventional Weapons Convention6 have

agreed to hold this year a four-day intergovernmental meeting to explore

questions related to lethal autonomous weapon systems with a view to po-

tentially drafting a Protocol VI to the Convention.7 Even the United States,

which is among the most technologically advanced States in this field, cur-

rently requires that such weapon systems “be designed to allow command-

ers and operators to exercise appropriate levels of judgment over the use of

3. Special Rapporteur on Extrajudicial, Summary or Arbitrary Executions, Rep ort, ¶

39, U.N. Doc. A/HRC/23/47 (Apr. 9, 2013) (by Christof Heyns) [hereinafter Heyns].

4. PETER W. SINGER, WIRED FOR WAR: THE ROBOTICS REVOLUTION AND CON-

FLICT IN THE 21ST CENTURY 124–27 (2009).

5. HUMAN RIGHTS WATCH, LOSING HUMANITY: THE CASE AGAINST KILLER RO-

BOTS (2012), available at http://www.hrw.org/reports/2012/11/19/losing-humanity; Eu-

ropean Parliament Resolution on the Use of Armed Dr ones ¶ H.2(d) (2014/2567(RSP))

(Feb. 25, 2014); Peter Asaro, On Banning Autonomous Weapon Systems: Human Rights, Automa-

tion, and the Dehumanization of Lethal Decision-Making, 94 INTERNATIONAL REVIEW OF THE

RED CROSS 687 (2012).

6. Convention on Prohibitions or Restrictions on the Use of C ertain Conventional

Weapons Which May be Deemed to be Excessively Injurious or to Have Indiscriminate

Effects, Oct. 10, 1980, 1342U.N.T.S. 137.

7. Meeting of the High Contracting Parties to the Conventio n on Prohibitions or Re-

strictions on the Use of Certain Conventional Weapons Which May Be Deemed to Be

Excessively Injurious or to Ha ve Indiscriminate Effects, Geneva, Switz., Nov. 14–15,

2013, Final Report, ¶¶ 18, 32, U.N. Doc. CCW.MSP/2013/10 (Dec. 16, 2013), available at

http://daccess-ods.un.org/TMP/4361535.01272202.html.

International Law Studies 2014

310

force,”

8 which means that it is not admissible for producers to program

machines that make final decisions as to targets against which force shall be

used.

It is perhaps because I have been confronted in actual armed conflicts

with so many violations committed by human beings, but inevitably never

with atrocities by robots9 (although admittedly, they did not exist in the

armed conflicts I witnessed), that my first feeling is not skepticism, but

hope for better respect of IHL.10 Only human beings can be inhuman and

only human beings can deliberately choose not to comply with the rules

they were instructed to follow. To me, it seems more reasonable to expect

(and to ensure) a person who devises and constructs an autonomous

weapon in a peaceful workplace to comply with IHL than a soldier on the

battlefield or in a hostile environment. A robot cannot hate, cannot fear,

cannot be hungry or tired and has no survival instinct.11 “Robots do not

rape.”12 They can sense more information simultaneously and process it

faster than a human being.13 As the weapons actually delivering kinetic

force become increasingly quicker and more complex, it may be that hu-

mans become simply too overwhelmed by information and the decisions

that must be taken to direct them.14 Human beings often kill others to

avoid being killed themselves. The robot can delay the use of force until

the last, most appropriate moment, when it has been established that the

target and the attack are legitimate. Certainly, there may be technical fail-

ures, but all those who drive cars and every traffic policeman know that

8. U.S. Department of Defense, DoD Directive 3000.09, Autonomy in Weapon Sys-

tems ¶ 3(a) (2012).

9. The terms “atrocity” or “violation,” like many other terms which describe human

behavior used in this article, are admittedly not appropriate for a machine.”

10. For a detailed, powerful, but not always entirely convincing plea, see RONALD C.

ARKIN, GOVERNING LETHAL BEHAVIOR IN AUTONOMOUS ROBOTS (2009). For a positive

assessment and call for regulation, see Kenneth Anderson & Matthew Waxman, Law and

Ethics for Autonomous Weapons Systems: Why a Ban Won’t Work and How the Laws of War Can ,

HOOVER INSTITUTE (Apr. 9, 2013), http://www.hoover.org/publications/monographs

/144241.

11. Ronald C. Arkin, Ethical Robots in Warfare, GEORGIA INSTITUTE OF TECHNOLOGY

(Jan. 20, 2009), http://www.cc.gatech.edu/ai/robot-lab/online-publications/arkin-re v.pdf;

Special Rapporteur on Extrajudicial, Summary or Arbitrary Executions, Interim Report, ¶ 20

(by Philip Alston), transmitted by Note of the Secretary-General, U.N. Doc. A/65/321 (Aug. 23,

2010) [hereinafter Alston].

12. Heyns, supra note 3, ¶ 54.

13. SINGER, supra note 4, at 127.

14. Arkin, supra note 11; SINGER, supra note 4, at 128.