ABSTRACT
With the development of robots, robotics and technology, it is expected that robots will become more prevalent in our lives. Increased use of robots will bring legal problems. The interaction of robots with each other and the interaction of robots with people can have consequences in terms of the law. This article aims to discuss the debates regarding legal personality and the responsibilities of artificial intelligence systems, and to identify where legislation needs to be developed in this area. First of all, we will consider what artificial intelligence is. Following this, legal developments in America, Europe and Turkey related to artificial intelligence will be discussed. Further, the legal identity of robots with artificial intelligence will be examined and the results of their actions will be evaluated in the light of the studies carried out around the world. In this study, we will look at American Law, European Law and Turkish Law, and legal regulations related to legal liability of artificial intelligence and robotics.
I. WHAT IS ARTIFICIAL INTELLIGENCE?
Although there are many definitions of artificial intelligence, the most common one is: “the high mental ability of a computer or a computer controlled machine to perform tasks related to high mental processes, such as reasoning, deducing, generalizing and learning from past experiences, which are often considered to be human-specific attributes2.” Artificial intelligence is the name given to the thinking initiative of a programmed computer.
Work on Artificial Intelligence began as far back as the development of the first modern computer. The father of the field was Alan Mathison Turing, who introduced the idea of thinking machines3. In 1943, the computer and the concept of artificial intelligence were formed out of devices produced for the needs of cryptoanalysis during the Second World War. Today witnesses the fastest progress in artificial intelligence and robots with the increase in processor capacity and processable data size, the impact of new algorithms, and the increase in investment. We are now using artificial intelligence at every point of our lives. Artificial intelligence is being used to find solutions for traffic problems, which are of particular significance in big cities. Navigation applications, which have become a part of our lives, show the fastest route, road works, accidents and similar traffic conditions. In air travel, for many years aircraft have been under human control for on average only seven minutes, most of the flight being on automatic pilot4. Research on autonomous vehicles is continuing rapidly around the world and signs suggest it will start to be used in the near future5. In addition to this, artificial intelligence is used in the field of e-mail, classifying our mails according to our needs. In the field of banking, credit evaluation, trading with mobile checks, and controls for fraud are all done through artificial intelligence. In Turkey, banking transactions are being done through online banking, rather than the number of branches increasing every year6. In addition to these examples, almost all applications used in social media and the entertainment industry use artificial intelligence, serving our interests. The smartphones we use have become more simplified in foresight that we fear as people in the future. Our smartphone assistants answer our questions, fulfill our instructions and make recommendations according to our interests. Unfortunately, the measures we take against the problems caused by artificial intelligence do not seem to be enough7. Since artificial intelligence is still human-supported, the regulations regarding the problems that artificial intelligence will create in the legal field are an area where the existing legal rules are insufficient in terms of the speed of development of technology and the issues it brings up. America, the European Union and Japan have raised the issue.8 Different countries have begun research into the influence artificial intelligence has on the law and into legal solutions to the problems of artificial intelligence9.
In Turkey, because of the low rate of development of artificial intelligence technology and the fact that it is limited to robots with non-autonomous artificial intelligence, no legal regulation has been made regarding the legal liability of artificial intelligence. This article will first examine the legal regulations related to the legal liability of artificial intelligence in America and Europe, and then examine what Turkish legal legislation can be applied to any problems of legality caused by artificial intelligence.
II. LEGAL REGULATIONS ON ARTIFICIAL INTELLIGENCE
A. Legal Regulations in Turkish Law on Artificial Intelligence and Robots
In Turkish Law, there is no regulation specific to robots, but the most recent regulation does issue instructions for uncrewed aerial vehicles. This instruction was prepared under the Law on the Organization and Duties of the General Directorate of Civil Aviation dated 18/11/2005 and the Turkish Civil Aviation Law No. 2920 dated 14.10.1983.
The scope of the instruction covers uncrewed aerial vehicles (UAV), operators of these vehicles, personnel employed and air traffic that flies in Turkish Airspace. Included in the instruction documents are requirements for keeping the UAVs and records. An Unmanned Aerial Vehicle is defined as an air vehicle that is controlled by a UAV pilot with a control link or whose autonomous operation is planned by a UAV pilot, except for model planes used for sports or recreational purposes, capable of flight without human power under the Instruction article 4/1-(ş). In this definition, attention must be given to the definition of UAVs as “controlled by a pilot.” Thus, the UAV cannot be considered independent of the owner or user. In this context, in case of any violation of security and confidentiality by a UAV, the UAV operator is responsible as the real or legal person10. Although there are no other regulations on artificial intelligence and robots in Turkish law, it is expected that legal regulations will be needed in the area of criminal law and obligations law11.
B. Legal Regulations in American and European Law on Artificial Intelligence and Robots
The legal liability of artificial intelligence was first discussed in 1979 in a case in Michigan, U.S.A.12 In a lawsuit, a jury ordered the payment of $10 million to the family of a person killed by a robot. The first legal arrangement was the law on driverless cars in the state of Nevada in 2011.13 Then the Federal Aviation Administration established that unmanned aerial vehicles are aircraft, and this has limited the controversy14. Although there is no legal regulation on the status of robots in American Law, Calo, who has researched Robot Law, says judges in American courts view artificial intelligence robots as independent controlled machines. Objections to this approach are concerned about the independent identification of artificial intelligence robots15. Although the courts continue to make decisions based on the jurisprudence they have determined, there is continuing work in the U.S.A. on making changes to the laws on artificial intelligence and robots; it is a continually evolving field16.
On 20 January 2015, the European Parliament’s Legal Affairs Committee (JURI) established a Working Group on legal issues related to the development of Robotics and Artificial Intelligence, with the primary intention of preparing the relevant legal rules. In their report, 38 general principles regarding the development of robots and artificial intelligence were considered. In the last section of the report, there are some suggestions for robot designers and users17. The report recommends that robot designers take into account the value of dignity, freedom and justice before, during and after the design, development and presentation of such technologies; define reliable principles for robot design; develop design features to ensure that private information is safe and appropriate; get a positive opinion from an ethics committee before testing a robot in a real environment or before incorporating people into the design and development procedures; and design robots compatible with legal principles. The report also discusses the rights of robot users. According to the report, users have the right to use robots in a risk-free environment and without fear, and users have the right to use robots according to the purpose of existence.
In addition to the working group mentioned above, “SPARC” was established to maintain and improve the position of Europe in the field of robotics, and “euRobotics” was set up to provide solutions for problems in the field of robot law. “SPARC” identifies and carries out studies in areas where robots can be used in human life, for example, manufacturing, health, trade, logistics and the military18. RoboLaw also carried out a study of robot law, which was completed in 2014. The primary objective of the RoboLaw project was to understand the legal and ethical impacts of emerging robotic technologies and to determine whether the existing legal frameworks are adequate and feasible in the light of the development and rapid proliferation of robotic technologies19.
The European Union’s European Robotics Research Network, also known as EURON, aims to create a roadmap for robotics in the next two decades. The network was established as a joint research center due to the inadequacy of European legal research on artificial intelligence and robots compared to that in Asian countries and America. They aim to continue their studies in all areas of robot technology, focussing on system integration, virtual learning and legal studies20. While this work on robot law in America and the European Union continues, one of the most critical legal developments in this field is undoubtedly the study on applying the status of electronic person to robots21
III. LEGAL STATUS AND LEGAL LIABILITIES OF ARTIFICIAL INTELLIGENCE AND ROBOTS
As the speed of autonomous robotics and artificial intelligence technology increases daily, in cases of artificial intelligence and robots, it is becoming increasingly questionable whether artificial intelligence and robots will be awarded legal personality.
In order to explain the legal liability of robots, the RoboLaw Project, which began in 2012, first grouped robots under three headings: autonomous robots, non-autonomous robots and intelligent robots. Autonomous robots are those that can make decisions by themselves in the outside world and implement these decisions without someone interferring22. In this context, robots that cannot be intercepted and can act without any intervention are called non-autonomous robots. Intelligent robots are described as robots that can act by adapting to their environment and/or interacting with external sources and adapting their behaviors as a result of this perception and interaction. Based on these categories, in order to be able to talk about the legal liability of artificial intelligence, artificial intelligence must acquire a legal personality. Given these defintions, it is clear that it is not possible to talk about the responsibilities of non-autonomous robots. Since these robots cannot make any decisions on their own, under liability law, robots cannot be regarded as having liability.
The European Parliament, in its motion on Civil Law Rules on Robots,23 emphasizes that in order to be defined and accepted as electronic persons, robots must be able to make intelligent, autonomous decisions, or if they can interact with third parties, they are responsible for their actions and only in this way can robots be given legal personality. The Parliament states that at such time as a robot cannot be controlled by another, it will be necessary to attribute legal personality to him. As robots are still controlled by humankind, they therefore cannot, for the time being, be afforded legal personality.
Although there are no legal regulations on artificial intelligence and robots in Turkish Law, it is possible to argue that solutions can be found by applying existing laws. Before looking at the provisions to be applied in Turkish Law, we focus on the legal personality.
It is important to determine the ability of robots to carry out actions and operations in order to be able to assign personal rights and define robots as a legal subject. In this context, in order to determine whether or not robots will be recognized as legal subjects in Turkish Law, it is useful to examine the issue in terms of the capacity to have rights and actions. According to Article 8 of the Turkish Civil Code No. 4721, “each person has a capacity of rights. Accordingly, all human beings are equal within the boundaries of the legal order to have rights and debts.” Article 9 and Article 10 of the Turkish Civil Code24 stipulate that a person who possesses the capacity to act (a person who has the capacity to discriminate and to act) has the right to be dealt with by his/her actions. Acting according to the wording of the law, a license can be defined as the right of a person to be entitled to his/her own will, to change these rights, to eliminate, to get into debt, to change debts, in other words, to be entitled to its own acts and to get into debt25. The concept of a person in law refers to assets that may have rights and debts. Being a human being is not a necessary condition for possessing legal personality – a concrete example of this is the legal person26. In other words, the legal order accepts two kinds of legal personality, one being a real person and the other being a legal person. Real people are just people. The legal person includes people’s communities and commodities that have been established in order to achieve a specific purpose and have the conditions required by the legal order. Because they have rights and liabilities, both the real and the legal person have many fundamental rights such as opening a case, filing a lawsuit, and the right of ownership. How can robots and users be held responsible for these actions when autonomous robots are not considered as persons in current law yet are intelligent learners and act on their own initiative?
The article points that can be overcome in the near future within the framework of the pace of development of artificial intelligence and robotics in Turkey will be made an observation and assessment of the legal troubles that can be experienced by the target. Artificial intelligence technology in Turkey, unfortunately, is a field study conducted over several years yet. Therefore, its development is slow and inadequate compared to other examples in the world. Because the development rate is low and the scope of such research is limited, it is not necessary to talk of autonomous robots in artificial intelligence and robotics technology and to make a new legal personality to robots and to make new laws mentioned in the liability of the autonomous robot. However, despite the possibility of legal problems related to intelligent robots, which we call semi-autonomous, in the Turkish Legislation, the regulations that can be applied by way of a comparative method in case of a legal problem related to semi-autonomous robots will be mentioned.
A. In Terms of Obligations Law
In Turkish legislation, the comparable provisions that can be used to determine the legal status of robots that use artificial intelligence are, firstly, the provisions for liability without fault. According to this legislation, responsibility lies with the employee, the legal person, the owner of the animal, the head of the household, the vehicle operator and a person under incapacity. Under this legislation, the person to be held responsible for an artificial intelligence robot is the owner or operator. These people will be held responsible even if there are no defects in the actions of the robots in their possession.
First of all, if it is necessary to evaluate within the scope of the Law of Obligations, it is possible to examine the cases of liability without fault in terms of enterprises and individuals using artificial intelligence robots. Given the low level of development in this field of technology in Turkey, the artificial intelligence used in the near future will be limited to robots with semi-autonomous artificial intelligence or intelligent robots. The development and use of fully autonomous robots that would require a separate regulation in Turkish Law is currently not being developed27.
The Law of Obligations addresses the responsibility of those who do not have the power to distinguish the “equity responsibility” set out in Article 65. The responsibility of those who do not have the power to discriminate is broadly kept. Accordingly, in cases where equity requires, it is possible to hold a person who has the proper relation of illegality, loss and irregularity and is responsible for the damage that is not defective, even if it is not defective28. In the case that artificial intelligence robots are held accountable for such a comparison because we are not talking about an autonomous robot, the damage to compensation law will be caused by the operator or owner of the robot.
The Code of Obligations regulates the care liability after the equity liability. Care liability is divided into three, namely the liability of the employee, the responsibility of the aminal owner and the responsibility of the building owner. Here, the responsibility for semi-autonomous or intelligent robots will be the responsibility of the employee to be balanced.
Article 66 of the Turkish Code Of Obligations stipulates that the “employee is obliged to prevent harm to others during the performance of the work given to him/ her.” Here, the operator of a semi-autonomous robot seems to be incapable of compensating the damages he incurred while AI was doing its job. In the continuation of the article, there is a proof of salvation that the employee will not be held responsible if he shows due diligence. In the case of the actions of semi-autonomous robots, presenting proof of salvation of the business owners or robot owners, as in the responsibility of the employee, will not be applicable to the robots, since the business owner who has fulfilled the care obligation would not be able to recourse to the semi-autonomous robot. However, if there is a fault in the software of the robot, it can be mentioned that the owner of the robot recourse to the designer of the robot.
Article 67 of the Code Of Obligations regulates the responsibility of the animal owner29. In this article, the person who takes care of the animal is obliged to eliminate the damage caused by the animal and can be relieved of this responsibility if he/ she proves that he/she has taken care to prevent the damage from being born. If the animal is intimidated by someone or another animal, the person holding the animal has the right of recourse to the person or the owner of the animal. In case semi-autonomous robots are used in our homes, this article may be applied by comparison. Electronic devices with software are already in our homes. If these electronic devices are semi-autonomous and cause damage as a result of their actions, the owner of the semi-autonomous robot will have to remedy this responsibility.
Another type of responsibility under the Code of Obligations is the liability of danger30. The liability of the danger set out in Article 71 of the Turkish Code of Obligations; it regulates that the owner and, if there is, the operator shall be jointly liable for the damages arising from the activities of the enterprise which is a significant danger. Firms using artificial intelligence or semi-autonomous robots in their enterprises can be considered as a liability in this way because they cannot foresee what this new technology will lead to. Article 36931 of the Turkish Civil Code entitled the liability of the head of the family, can be applied if a robot used in the household is considered a family member. According to this provision: “The head of the household is responsible for damages caused by the household, the minor, the person who has mental illness or the mentally ill unless the patient proves that he is unable to prevent the damage from occurring, even if he/she observes him/her carefully. The head of the household is obliged to take necessary measures to ensure that those who have mental illness or mental weakness from their households do not put themselves or others in danger or harm. It shall ask the competent authority to take the necessary measures under its obligation.” The fact that semi-autonomous robots are in the status of the person under incapacity appeal is a status that other countries have included in their discussions. Evidence of salvation is presented to the head of the household if the necessary care is shown, as in the responsibility of the employee. If the substance is applied in comparison with semi-autonomous robots, it is more likely that the proof of salvation will not be entirely free of liability but a reason for a reduction in compensation.
B. In Terms of Criminal Law
As in the assessment of the law of obligations, an assessment will be made for robots with semi-autonomous artificial intelligence which will be used in the near future. In this sense, a scenario in which the robot with artificial intelligence comes to mind first is not responsible for the criminal law. Nevertheless, it would be necessary to examine criminal liability in terms of several possibilities.
1. Use As a Tool
The first possibility is to commit a crime under the penal code by the use of the robot as a human tool. The first possibility is that the programmer is responsible for using a robot as a tool. This is the first solution that comes to mind. The robot programmer has caused the robot to commit a crime because of an error in the software, but in order to be able to talk about such a situation, the programmer must have a flawed behavior33. Another example of the use of robots as a tool is that the user is responsible. The user’s responsibility will be especially the case where it is intended to damage the use of artificial intelligence as a weapon. Article 6 of the Turkish Penal Code defined the weapon. According to this definition of weapons; 1. firearms, 2. explosives, 3. all kinds of cutting, penetrating or bruising tools made for use in defense and attack, 4. for the purpose of attack and defense, other things which are de facto suitable for attack and defense, burner, corrosive, hurtful Nuclear, radioactive, chemical, biological substances that cause persistent, persistent, toxic, persistent disease are considered weapons. It is, of course, possible to consider a semi-autonomous or non-autonomous robot as a weapon in the sense of Article 6 of the Turkish Penal Code. Another possibility is the liability of someone else’s action. In this example, the semi-autonomous robot is now used as more than a tool. In this example, the provision in the criminal law is mentioned in the same statute as the people who lack the power of appeal under the law of obligations. In this scenario, which is considered as a criminal offense of a person whose defect ability has been eliminated, the indirect agency will come into play. Indirect perpetrator as regulated by Article 37/2 of the Turkish Penal Code, “the punishment of the person who uses another person as a tool for the crime is increased by one third to half.” In the previous examples, the artificial intelligence robot is used as a simple tool, but in this example, it is like a human being used as a tool. In this case, the perpetrator will be the programmer or the user. In this example, robots are more likely to be autonomous.
2. Liability of Negligence
Negligence is the unforeseeable result of an act which is specified in the statutory definition of the crime. In order to be able to talk about the liability of the negligence, artificial intelligence should have caused an unlawful result based on the accumulated experience or knowledge34. Examples are semi-autonomous vehicles. The fact that the focus on autonomous vehicles in artificial intelligence studies and even the use in some countries may indicate that semi-autonomous robots will exist primarily in motor vehicles. In this context, although it is not yet carried out in our country, one of the artificial intelligence that we will encounter in the near future will be autonomous or semi-autonomous motor vehicles. Without the need for these vehicles, the road, traffic flow and the perimeter of the environment can be detected by moving. For example, if the drivers of such vehicles do not allow this and cause an accident while attempting to deactivate artificial intelligence, then the responsibility will be on the user or the programmer. This time, however, the form of liability changes because the example here is that the user and the programmer do not act with the intent of committing a crime but do not show due diligence.
IV. CONCLUSION AND EVALUATION
Although robots with artificial intelligence have not yet taken place in our lives as robots and human beings, we use artificial intelligence in many areas. The rules of law aim to ensure the social order and what is still to be done in the face of the behavior of an innovation such as artificial intelligence, which is not the subject or object of the rules, is against the legal rules. According to our assessment, due to the lack of sufficient rules in this area, artificial intelligence designers need to design in accordance with the current legislation, fundamental rights and freedoms. In the meantime, taking into account the use of artificial intelligence and robot technology in our country and the developments showed, it is likely that the regulations in the field of obligations law and criminal law can be applied to the disputes by comparison with the judges. If new developments are insufficient in terms of robots with artificial intelligence, it is possible to talk about new robot legislation.
BIBLIOGRAPHY
ARMAĞAN EBRU YUKSEL BOZKURT, Robot Law, Justice Academy of Turkey 2017.
ERICA PALMERINI, Regulating Emerging Robotic Technologies in Europe: Robotics facing Law and Ethics, Robolaw, Guidelines on Regulating Robotics, 2012.
ESRA DEMIR, Robot Law, Istanbul Bilgi University, Institute of Social Sciences, Information and Technology Law, 2017.
ERNEST DAVIS, Artificial Intelligence, A Modern Approach, PearsonEducation,Inc., Upper Saddle River, New Jersey 2010.
FEDERAL AVIATION ADMINISTRATION, Unsafe/ Unauthorized UAS – What Can You Do?, 2018.
GABRIAL HALEVY, “When robots kill : artificial intelligence under criminal law / Northeastern University Press, 2013.
GÖKHAN ANTALYA, Genel General Provisions of Law of Obligations, Adalet Yayınevi, 2008.
HALIT KORKUSUZ, The Place of Danger Responsibility in Our Law, Journal of Dicle University Faculty of Law, Volume: 15-16, Issue: 22-23-24-25, Year: 2010-2011.
HALLEVY GABRIEL, “The Criminal Liability of Artificial Intelligence Entities – From Science Fiction to Legal Social Control”, 2010.
MEHMET MEVLÜTOGLU, Robotic Technologies Sector Report, Robotics Automation and Artificial Intelligence 2015.
NEIL RICHARDS, “How should the law think about robots?”, Washington University School of Law; Yale Information Society Project; Stanford Center for Internet and Society 2013.
RICHARDS, NEIL M, “How should the law think about robots?”, Washington University School of Law; Yale Information Society Project; Stanford Center for Internet and Society 2013.
ROBERT, SPARROW, “Killer Robots”, School of Philosophy and Bioethics, Faculty of Arts, Monash University, 2007.
ROBERT, SAWYER, “Robot Ethics”, Sciencemag.org., 2007.
RYAN, CALO, “Robots in American Law”, University of Washington School of Law, Legal Studies Research Paper, 2016.
SHAI BEN-DAVID, SHAI SHALEV-SHWARTZ, Understanding Machine Learning: From Theory to Algorithms, Cambridge University Press 2014.
TEYFIK KARADUMAN, Artificial Intelligence Application Areas, Gazi University Institue of Informatics. A.B.D.
TUBA PEKMEZ, Criminal Liability fort he Use of Autonomous Vehicles, 2018.
FOOTNOTE
2 Teyfik Karaduman, Artificial Intelligence Application Areas, Gazi University Institue of Informatics. A.B.D.
3 Ernest Davis, Artificial Intelligence, A Modern Approach, PearsonEducation,Inc., Upper Saddle River, New Jersey 2010.
4 http://boardinginfo.com/otopilot-nedir/.
5 Tuba Pekmez, Criminal Liability fort he Use of Autonomous Vehicles, 2018.
6 Bank Association of Turkey, Digital Internet and Mobile Banking Statistics 2018
7 Hallevy Gabriel, “The Criminal Liability of Artificial Intelligence Entities – From Science Fiction to Legal Social Control”, 2010.
8 Ebru Yuksel Bozkurt, Robot Law, Justice Academy of Turkey 2017.
9 Richards, Neil M, “How should the law think about robots?”, Washington University School of Law; Yale Information Society Project; Stanford Center for Internet and Society 2013.
10 Uncrewed Aerial Vehicles Instruction, article 6/2.
11 Ebru Yüksel Bozkurt, “Robot Law” Justice Academy of Turkey, 2017.
12 Robert, Sawyer, “Robot Ethics”, Sciencemag.org., 2007.
13 Richards, Neil M, “How should the law think about robots?”, Washington University School of Law; Yale Information Society Project; Stanford Center for Internet and Society 2013.
14 Federal Aviation Administration, Unsafe/Unauthorized UAS – What Can You Do?, 2018.
15 Ryan, Calo, “Robots in American Law”, University of Washington School of Law, Legal Studies Research Paper, 2016.
16 US National Science and Technology Council, Preparing fort he Future of Artificial Intelligence, 2016.
17 European Parliament, Committee on Legal Affairs, PR\1095387EN, Draft Report 2019.
18 https://eu-robotics.net/sparc/
19 http://www.robolaw.eu/
20 Esra Demir, Robot Law, Istanbul Bilgi University, Institute of Social Sciences, Information and Technology Law, 2017.
21 Erica Palmerini, Regulating Emerging Robotic Technologies in Europe: Robotics facing Law and Ethics, Robolaw, Guidelines on Regulating Robotics, 2012.
22 Erica Palmerini, Regulating Emerging Robotic Technologies in Europe: Robotics facing Law and Ethics, Robolaw, Guidelines on Regulating Robotics, 2012.
23 Erica Palmerini, Regulating Emerging Robotic Technologies in Europe: Robotics facing Law and Ethics, Robolaw, Guidelines on Regulating Robotics, 2012.
24 Turkisch Civil Code, Article 9: The person who has the capacity to act can be entitled to and be indebted by his own actions.
25 Turkish Civil Code, Article 10: Any adult person who has the power discriminate and has no restriction has the capacity to act.
26 Turkish Civil Code, Article 48: Legal persons shall be entitled to all rights and debts except those that are dependent on human-specific qualities such as sex, age, kinship
27 Mehmet Mevlütoglu, Robotic Technologies Sector Report, Robotics Automation and Artificial Intelligence 2015.
28 Gokhan Antalya, Genel General Provisions of Law of Obligations, Adalet Yayınevi, 2008.
29 Turkish Code of Obligations, art 67, Responsibilty of Animal Owners.
30 Halit Korkusuz, The Place of Danger Responsibility in Our Law, Journal of Dicle University Faculty of Law, Volume: 15-16, Issue: 22-23-24- 25, Year: 2010-2011.
31 Turkish Civil Code, Article 369, House Order.
33 Robert, Sparrow, “Killer Robots”, School of Philosophy and Bioethics, Faculty of Arts, Monash University, 2007.
34 Gabrial Halevy, “When robots kill: artificial intelligence under criminal law / Northeastern University Press, 2013.
