Is it possible to grant legal personality to artificial intelligence software systems?

Authors – Paulius Čerka, Jurgita Grigienė, Gintarė Sirbikytė.

The purpose of this paper is to determine whether Systems of Artificial Intelligence (SAI) can be deemed subjects of law. This aim is formulated according to the technical capabilities integrated in SAI and the SAI’s ability to interact independently with other legal subjects. SAI features, such as direct connection with intellectual skills, the ability to understand, learn and make autonomous decisions may cause situations where autonomous systems based on AI will make decisions which will be in the best interests of individuals, even though conflicting with the user’s own will.

To consider the possibility of SAI being recognized as possessing legal personality, we analyse the concept and features of SAI and define its operating principles. We give hypothetical examples to demonstrate the necessity of SAIs being recognized as such. The paper undertakes legal personality analysis of SAI performed: (i) using the philosophical and legal concepts of a subject (person); (ii) discussing artificial (unnatural subjects of law) as an alternative to the recognition of legal personality of SAI; (iii) using elements of legal personality set for natural and legal persons.

The analysis leads to the conclusion that the scope of SAI rights and obligations will not necessarily be the same as the scope of rights and obligations of other subjects of law. Thus, SAI could only have rights and obligations that are strictly defined by legislators. This conclusion suggests that the result of this paper may be its use in further research defining the scope of SAI rights and obligations.

Introduction

“Can a machine think?” The question attracting more and more attention of scientists and practitioners was raised back in 1950 by Alan Turing1  who thereby set a direction for the discourse on Artificial Intelligence (AI). AI is rather new discipline, which has no single definition yet. The concept of AI was first mentioned in 19562; the system is often defined as artificially developed intelligence related to rapidly developing technologies, which enable computers to operate intelligently, i.e. in a human-like manner. Systems of Artificial Intelligence (SAI) are different from other regular computer algorithms (programs) due to their uniqueness, since they are able to learn independently, gather experience and come up with different solutions based on the analysis of various situations independently of the will of their developer (programmer), i.e. SAI are able to operate autonomously rather than automatically. 

Today, there is a great variety of information technologies based on the operating principle of SAI, for example, Google Self-Driving Cars, autopilots controlling airplanes, digital assistants such as Siri, Cortana and Google Now, robot nurses, mind-controlled Google smart glasses, etc. These and other technologies are known worldwide and their capacity as well as use is rapidly growing. As the use of technologies based on AI becomes more and more widespread, the number of associated incidents grows as well. For example: (i) in 1.7 million miles of travel, Google Self-Driving Car had 11 accidents resulting in damage; (ii) speech recognition software could become a contributory factor to car accidents5; (iii) robot nurses reminding patients to take their medicines fail to ensure that the medicines are actually taken, which may lead to the patient’s death.

These specific examples show that SAI are not mere science fiction. Information technology innovation based on SAI and the above examples allow stating that SAI are not mere objects, operation of which is influenced by others. SAI act like entities. Regardless of the exceptional operating principle of such systems, none of the legal systems has recognized SAI subjects of law so far. However, is such legal status of SAI only a temporary attribute, which should change in time? Is it possible to grant legal personality to a System of Artificial Intelligence? Even though the problem at hand has already been addressed before, it has attracted greater (proper) attention only in the recent years. 

It should be noted that SAI that are capable of learning and making decisions independently can make the lives of people easier, but failure to manage such technology can lead to major existential threats. Bill Gates claims that after a few decades, SAI and the level of their intelligence will lead to major concerns; therefore, it is necessary properly to prepare ourselves. As the role of SAI in our daily lives becomes more and more important, we encounter various challenges: moral, ethical issues and problems. Legal regulation and legal system itself are not an exception. Discussions on the status of SAI that are able to make more and more complicated decisions independently in the legal systems of countries become increasingly frequent and extensive at the academic level, in the political arena of various countries as well as in the context of its shaping, and in the society.

The ability of SAI to learn from their own personal experience leads to independent conclusions and autonomous decision-making, i.e. what can lead them to their legal personality. Due to their ability to make decisions independently, technologies based on such systems like Machine Learning, Expert Systems or Neural Networks can no longer be treated as objects. Therefore, the aim of the paper is to determine whether SAI can be deemed subjects of law. The object of the paper is legal personality of SAI and the methods of research are information collection, systematizing, generalizing, valuation, comparison, analysis of scientific literature, synthesis and deduction. 


Download full article

Go back