Autonomous Robots and the Law

Automatic machines have been around in people’s fantasy or reality since ancient times. They are described in myths, such as that of Hephaestus (famous for his craftsmanship and for creating mechanical servants[1]), and in more technical manuscripts, such as the ones of Hero of Alexandria (who described a wide range of machines and automatons[2]). The history of those mechanical devices is not confined to Western stories and writings – references to a piece of work that ‘walked with rapid strides, moving its head up and down, so that anyone would have taken it for a living human being’ can be found in Chinese literature from 500 BC[3] and in 1206 AD the Arab inventor Ibn al-Razzaz al-Jazari published The Book of Knowledge of Ingenious Mechanical Devices.[4] These ideas and constructions have been followed by innumerable working creations and fictional characters, including the ones in the Czech writer Karel Èapek’s play R.U.R. (Rossum’s Universal Robots), which premiered in 1921 and introduced the word ‘robot’ (literally meaning work or labour but normally used for servant labour). But it can be argued that the idea of a future where the human environment was populated with robots was made mainstream by modern science fiction films. Since then people have not been wondering whether robots will be a common feature of their (or their children’s) future, but when that will materialize. In a sense, it is possible to quote an Argentine rock band and say that ‘the future has arrived a long time ago’[5].

Living with Robots

Today robots are widely used in factories and, if we understand the word in a broad sense, we have many of them at home. In this broader sense, most machines will be robots, since they can be used to carry out tasks replacing some human work – on that basis even a food blender is a robot. However, it is clear that machines that perform such simple domestic tasks don’t conform to the idea that most people have about what a robot is and, accordingly, a working definition could identify a robot as a machine or device that is able to perform different and variable tasks, can sense its environment, interacts with it and can also perform coordinated movements. There is an estimate of more than one million robots in the world and the number is expected to rise by more than 4% annually.[6] There are even those than can perform certain tasks better than humans, like the Da Vinci surgical system, and others that carry out tasks that human could do but only in a less efficient manner or while putting themselves at risk, like those used to build cars or help the military in war zones. In general terms, the ultimate end of robotics could be seen as the production of machines that can carry out the requested task in changing environments and without continuous human guidance, vg autonomous robots or what are also called Next Generation Robots.

Completely autonomous robots are not the exclusive domain of science fiction and there have been certain technological breakthroughs raising the expectation that the day when robots become a distinctive group cohabiting with humans is fast approaching. In 2005 a team of scientist at Cornell University demonstrated a self-replicating robot that could lead to machines that clone themselves[7] and the advances in neurology have led to substantive advances in the areas of artificial intelligence. While artificial intelligence and robotics are separate fields of research, the creation of fully autonomous robots depends on their capacity to act intelligently and some are of the belief that due to Moore’s law[8] and some theories of cognitive science, the date of such a leap forward can be predicted with real accuracy.[9] Although not exactly autonomous, there are already several examples of robots that increasingly take decisions based on few original orders and can adapt to changing environments. Probably due to the size of the research and development budgets, and fuelled by the need to minimize casualties, many of those examples can be found in military uses of partially autonomous robots. The SGR-A1 produced by Samsung for the South Korean military is a smart, unmanned gun tower, intended to work on the border between North and South Korea, which has the capability to fire on its own. If deployed, this system would be the first robot to autonomously shoot at a target and perhaps kill a human being.

Leaving apart the practical considerations that such a development would encompass and avoiding the apocalyptic view that sees robots becoming self aware and self reproductive and taking over society for their benefit (the future presented in the Terminator saga where machines are stronger and smarter than humans and are determined to wipe humanity off the face of the earth), it is important to recognize that a reality where interaction with robots becomes the norm is not so distant in our future. The impact that such reality will have on the legal system of every jurisdiction could be broader and deeper than what a superficial analysis would show. Governments have realized this situation and several have established task forces like the Japanese Robot Policy Committee (established by the Ministry of Economy, Trade and Industry) or have ordered work tending to assess the consequences for society, including the legal system. In the UK, the Office of Science and Innovation commissioned a report about the future of robots in society, but the result was a piece that unfortunately wandered into some esoteric issues of robo-rights and pensions for robots instead of focusing on the real effect that truly autonomous robots would have for humans, like the centuries-old issues of civil liability and criminal culpability for the actions of these intelligent machines.

There is a growing body of research devoted to issues relate to the use of autonomous robots in war zones and how they could be designed to comply with the Laws of War and the established Rules of Engagement. But there are other areas where establishing issues of liability could be more problematic. The deployment of robots in unknown environments where humans or animals react in unexpected ways could lead to undesired effects resulting in damages or a crime and this situation is so regardless of the level of autonomy that the robot in question has, because it is accepted that non-autonomous robots could behave in unexpected ways when submerged in new environments and, in the case of the still fictional fully autonomous robot, it would be the robot which is the one consciously making the decision of what to do under those circumstances.[10]

Robots in a Sexual Role

Sex robots or sexbots are one of the areas where these problems are bound to arise in way that challenge legal concepts and principles. If we are going to juggle the problems that robots set for the law, we may as well do it in an area that captures our attention.

 

Some people tend to have an instinctive negative reaction on the issue of humans having sex with robots, but there are some signs pointing towards a situation where sex robots become common for many people. The back cover of David Levy’s book ‘Love and Sex with Robots’[11] refers that ‘[f[rom Pygmalion falling for his chiselled Galatea to the man-meets-machine fiction of Philip K Dick, humans have long been enthralled by the possibilities of emotional relationships with their technological creations’ and the text makes some valid points about sex robots providing companionship and satisfaction for those who have no time, will or luck to find them in real partners. It is not necessary to agree, or even to consider the possibility of Levy’s theories being true, to realize that, either based on the Marlow’s hierarchy of needs, the disputed Freudian theories of sexual desire as the primary motivational energy for human activity or some market facts, sexual activity has played a pre-eminent role in what humans do and that the search for physical and spiritual company occupies important parts of people’s time and budgets. It could be argued that every technological development has given place to a market where that technology has been used for sexual purposes. Human genitalia carved in wood dating from the Upper Palaeolithic was found in excavations and all other ancient technologies, from pottery to work in leather, had been used to such effect. From the communications technologies point of view, the printing press opened the door for text and graphic representation of explicit sex, photography and cinema created one of the biggest creative pornographic industries, telephones led to phone sex and pornography engenders 12% of all Web sites, 25% of search engine requests and 35% of downloads.[12] It is no surprise that in the area of robotics, sex has also been one of the driving factors, from the non-robotic basic sex doll to the already existing sex bots, all pointing to a future where autonomous robots will be used as sexual partners, and the legal implications of that reality would deserve not an article but a full encyclopaedia. In Japan the Repliee Q1Expo is a female robot that has silicon skin, 42 compressed air-controlled actuators and is controlled by a computer which enables it to mimic the movements of a real human, including fluttering her eyelids and breathing.

Under the current stage of technological development of robots, most legal issues are reduced to product liability, negligence and, in some cases, criminal responsibility, but always falling on the programmer, manufacturer, seller or owner of the machine. However, when dealing with sex robots of the Next Generation Robots, the legal issues may not be so straightforward or the answer to them may not be in any way clear, assuming that the relevant answer exists at all.

Sex Robots and Rape

Taking into account the aforementioned about both autonomous and non-autonomous robots, the situation could exist where a sex robot engages in sexual activity with a person against the wishes of that person. In the case of non-autonomous robots, it was previously stated that they tend to act in unpredictable manner when facing new situations and one could expect that many unprogrammed, new and unexpected situations can arise in sexual encounters, including unrecognized orders or unusual tones of voice, which could disorientate the robot and lead it towards forcing upon the owner (now victim) non-agreed sexual intercourse. In the case of future autonomous robots with self-awareness, the robot can decide to ignore the will of the victim or may understand a negative response as a positive encouraging one. Under the Sexual Offences Act 2003, rape is defined as a ‘person’ committing an offence and the relevant subsection for the current analysis is the requirement of reasonableness in relation to the defendant’s belief in the victim’s consent. The first, and probably insurmountable, hurdle to build a case of rape in such a situation would be the requirement of having a person commit the actus reus, which might itself suffice to leave the issue and start writing about the latest Facebook story. However, these situations are likely to start happening in a not so distant future and the normal common-law approach to wait and see may not be the most advisable one here.

A probably uncomplicated solution would be to treat autonomous robots as persons, which would not be as far-fetched and extravagant as it may sound. Most legal systems have in one way or another given legal personhood to non-humans, eg corporations can be held liable in civil and criminal cases. It could be argued that the situations differ radically between a robot forcing sexual intercourse with a human and a gross breach of a relevant duty of care owed by the organisation to the deceased, as established by the Corporate Manslaughter and Corporate Homicide Act 2007, but they both constitute situations where the law creates a legal fiction of personhood to address a the result of an undesired effect of non-humans’ behaviour. The example of corporate liability does show that personhood can be created by law, and the creative capacity of legislators.

Assuming that the hurdle of personhood is overcome, the next issue would be the lack of consent and focusing on the requirement for the belief in the consent of the victim to be reasonable, as established by s 1(2) of the Sexual Offences Act 2003 and explained in ss 74, 75 and 76. In the case of the rebuttable presumption of lack of consent in s 75, a robot that is engaging in an act which is not desired by the human in question could create the circumstances stated in s 75(2)(a) and (c), since the victim would clearly have reasons to fear further violence towards him or her (the machine is already doing what is not supposed to do) and it is also likely that the robot, being a machine that can hug, could cause the victim to be detained. When dealing with the conclusive presumption in s 76, the situation may become more problematic because the robot may commit the offence by doing what it was programmed to do, like impersonating a third person loved by the victim because of therapeutical needs[13]. Could in this latter case the manufacturer or the programmer be accused of causing a person to engage in sexual activity without consent as per s. 4 of the Sexual Offences Act 2003? Furthermore, can the victim report the manufacturer and/or the programmer for assaulting him or her by penetration under s 2? The latter offence creates other possibilities when the owner of a sex robot is, say, showing the robot to a friend and, after indicating to the robot to show some preparatory movements, the robot fails or refuses to stop and forces intercourse on the friend. Would the owner be guilty of assault by penetration under s 2(a) or would the absence of intention exonerate him? It seems that the only clear answer is that there is no clarity in the answer or no answer at all.

Where under current or hypothetically reformed law a jury is able to find that a robot committed the actus reus of rape, or assault by penetration with some form of mens rea (or, better named, processor rea), what would be the punishment? Should a robot be punished? Punishment in criminal law assumes some form of ‘retributive practice whatever may be the further role of retribution as a (or the) justification or goal of punishment […] society needs the threat and practice of punishment’[14] to stigmatize and make the criminal reflect on his or her actions, all which seem unattainable or futile in the case of robots. If an autonomous robot commits a crime, a simple reprogramming could avoid further criminal activity unless the intrinsic characteristics of the design are the ones creating the possibility for the robot to perpetrate the offence, which would bring to mind the issues of product liability and, in some cases, some form of corporate crime.

Accordingly, it seems that under the present law most cases involving robots would have to be treated as being some kind of accident, leaving the victim with civil law as the only recourse to look for redress. There seems to be a strong case for some creative thinking in adapting the law to deal with issues like the one presented, but it is difficult to envisage a system of rules where criminal liability can be imposed upon robots in a manner that does not imply a revolutionary change in the way liability and causation are understood within the wider legal system.

Conclusion (or sort of)

Robots that can learn from past experiences and react to new environments are closer to fruition than people tend to imagine and, by definition, the days when one could predict exactly what a robot would do appear to be close to an end. It could be argued that the impact on the legal system cannot be properly assessed and that any conclusion would be pure speculation, but even the most basic analysis highlights the fact that, where a whole set of technology-based new entities enters into the system, the repercussions would be felt in almost every branch of the law and the change will have an impact on every legal principle.

That new scenario would impose a renovated stress on the discussion about whether the law should follow technology or try to regulate its behaviour. In the case of criminal liability, a simple analysis of sex robots and rape shows that current law and principles are ill-equipped to deal with further radical changes in technology, which could imply the need of a more proactive approach to legal developments.

 

Dr. Fernando Barrio is Senior Lecturer in Business Law and Programme Leader for the MA E-Business Regulation course at the London Metropolitan Business School.