Legal framework for AI applications

Artificial Intelligence Technology

Throughout history, humanity has been obsessed with developing devices that can mimic the cognitive processes of the human mind. The ultimate objectives have always been to improve human life, increase production in various industries, and reduce mistakes. For example, in 1872, Samuel Butler predicted that machines would be essential to human progress and the global advancement of prosperity.

Science fiction was the first genre to use artificial intelligence (AI). It sometimes emphasized the predicted bad features of artificial intelligence, depicting it as a ferocious enemy of humanity that seeks to control it, and at other times referred to it as a way to improve human life. A new chapter in the history of technology began in 2018 when artificial intelligence became a reality. Its integration has spread to other areas of daily life, such as traffic control, city navigation, and virtual assistants to assist with various operations. Today, artificial intelligence is widely recognized as an instrument that could boost society (Ghaleb, 2011).

Artificial intelligence has special characteristics, applications, advantages, and drawbacks like other concepts. Therefore, it is imperative to draft a law regulating all issues associated with it, such as its controls and legal liability. Therefore, the researcher has dedicated two sections to discussing these issues. The first deals with the AI concept, while the second section will discuss the legal nature of AI.

The Concept Of Artificial Intelligence

In April 2021, the European Commission proposed an artificial intelligence (AI) regulatory framework within the EU. The draft AI legislation represents the first attempt to pass a horizontal regulation for AI. The intended legal framework is centered on the particular application of AI systems and related risks. The Commission defined Artificial intelligence as “software that is developed with one or more of the techniques and approaches listed in Annex I and can, for a given set of human-defined objectives, generate outputs such as content, predictions, recommendations, or decisions influencing the environments they interact with.”

The World Intellectual Property Organization defines artificial intelligence as: “ a discipline of computer science that is aimed at developing machines and systems that can carry out tasks considered to require human intelligence. Machine learning and deep learning are two subsets of AI.” As for the legal aspect, Robert Little Dictionary defined artificial intelligence as: “a part of computer science that aims to simulate a cognitive ability to replace humans in performing appropriate functions in a specific context that require intelligence.” John McCarthy, the father of artificial intelligence, defined it as “the science and engineering of making intelligent machines.” And that “AI involves the creation of programs and machines that can perform tasks that, when done by humans, typically require intelligence. These tasks include learning, reasoning, problem-solving, perception, and language understanding.”

In Jordanian legislation, artificial intelligence was not specifically defined. However, Article (2) of the Electronic Transactions Law defines the electronic intermediary as “The electronic program or software used to carry out a procedure or to automatically respond to a procedure for the purpose of generating, sending, or receiving information message." This definition is deemed the closest to AI despite excluding certain characteristics and varieties. Regretfully, the legislator in Jordan is still ignorant of the gravity of the rapidly advancing applications of artificial intelligence and the appreciable rise in their use across all sectors of the economy, society, and agriculture.

Based on the definitions above, it is obvious that the variations in the concept of artificial intelligence result from the many lenses through which they are viewed. As a result, the researcher suggests the following definition of artificial intelligence: a scientific discipline focused on empowering robots to perform complex classification, specialized research, and analytical processes in various fields related to areas concerning human needs and developing machines capable of simulating the human mind, making decisions, solving problems, and providing solutions independently.

Legal Nature of Artificial Intelligence

Determining the legal nature of artificial intelligence requires defining its legal personhood. There has been a lot of controversy around the issue of granting artificial intelligence legal personality since it comes in various forms, such as machines and software operated by humans or independent robots that can make autonomous decisions (Salam, 2022). This variation makes it difficult to determine which law shall be adopted and applied to cases where risk results from machines operated by artificial intelligence.

Artificial intelligence was divided into three main categories (Salam, 2022) depending on the abilities they have such as:

The Jordanian legislator specifies in Articles (30-52) of the Civil Codes that legal personality is granted to a natural person eligible to acquire rights and obligations. It is also granted to judicial persons when they meet the specified conditions. In this context, we argue whether we can apply the rules pertaining to natural or legal personality to artificial intelligence or whether it needs special provisions; if so, a new legal amendment must be put in place (Al-Khouli, 2021).

The issue of granting legal personhood status to artificial intelligence has been widely debated. Some consider artificial intelligence to be a human product; therefore, it should be treated as property or under intellectual property rights laws, such as copyright, which should be applied to AI (Mohammad, 2021). Opposing opinions typically claim that artificial intelligence technology regularly outsmarts natural human intelligence, especially those who did not create it. This viewpoint holds that artificial intelligence (AI) arises as a separate, autonomous being that is not dependent on its human creators, making it eligible to have rights and obligations. The holder of this opinion argues that legal personhood should be granted to artificial intelligence (Badr, 2022).

The researcher claims no provision in the general rules states that legal personality is granted exclusively to natural and legal persons. Therefore, nothing is preventing artificial intelligence from being granted legal personality. However, the recognition of artificial intelligence as a legal personality as an independent entity in itself will be reflected in the prevailing legal system, as this will result in the artificial intelligence having its own independent financial liability under which it can acquire rights and bear obligations.

In this regard, the researcher suggests that legislators should urgently specify special regulations clarifying the legal personality of AI, especially in light of the tremendous development of artificial intelligence.

Legal Liability Of Artificial Intelligence

Artificial intelligence systems of all types have become a reality in our lives, and their applications are used to perform various tasks, whether in the educational, health, economic, or security fields. Since AI performs something physical, the likelihood of making an error is anticipated; therefore, whoever is responsible for this error must bear all liability and make good for any resulting damages. In the next chapter, we will discuss the civil and administrative liability resulting from using artificial intelligence.

Civil Liability Resulting From Utilizing Artificial Intelligence

In Jordan, there is a dramatic gap between the slow pace of making regulations pertaining to artificial intelligence, specifically regarding legal personhood, and the breathbreaking development of artificial intelligence applications in all aspects of life. Therefore, it is necessary to establish legal texts defining the scope of AI responsibility and provide texts protecting the rights of AI and protecting individuals from harm that may be inflicted on them by AI.

In this context, two theories have emerged regarding determining the responsibility of artificial intelligence: the traditional theory and the modern theory (Salam, 2022). The traditional theory claims that artificial intelligence may take on the nature of a physical thing directed by humans. Based on this opinion, the person who owns it bears full responsibility for guarding and protecting it and ensuring that it does not inflict any harm to others (Al-Khatib, 2018). For instance, Article (291) of the Jordanian civil law has stipulated that “Any person who has things under his control which require special care to prevent their causing damage or mechanical equipment, shall be liable for any harm done by such things or equipment, save to the extent that damage could not have been averted.” As for artificial intelligence machines and tools that are independent and capable of making decisions, some jurists believe that the one who ultimately bears responsibility for the damages resulting from it is the producer who created and programmed their software in the first instance (Al-Dahiyat, 2019).

Regarding protecting AI, Article (3/B/8) of the Copyright Protection Law No. (22) Of (1992) and its amendments stated that “The protection under this law includes Computer programs whether in the source language or machine language.”

Administrative Liability Resulting From The Utilization of Artificial Intelligence

The Supreme Court of Justice defined the administrative decision as: “a final legal act issued by the unilateral and binding will of the public administration, with the authority it has under the laws and regulations, and in the form required by the law, with the intention of establishing, amending, or abolishing a specific legal position for the public good.”

The technology revolution has greatly developed the administration sector, just like in other fields. Artificial intelligence is now a part of the administration process and may do many administrative tasks, including independent decision-making. Artificial intelligence has undoubtedly made significant progress in this area, as seen by its ability to streamline and reduce administrative procedures while guaranteeing prompt and efficient responses. Electronic administrative decision-making, which some jurists have described as “the responsible authorities' use of information systems to adopt one alternative from among the alternatives presented,” is one of the most notable examples of this evolution (Hosni, 2008).

However, this development raises crucial issues pertaining to the legal framework of the administration's responsibility for the actions and decisions carried out by artificial intelligence instead of the public employee.

From a technical standpoint, it is clear that artificial intelligence systems used in the administrative domain depend on a human operator who is skilled in artificial intelligence and provides the first input of data and information. Then, the system does all the instructions independently without needing a human operator. The next steps in this technological revolution are language comprehension, sound and visual fingerprint identification, and the ability to comprehend and use knowledge. The system gathers data from established laws, concepts, theories, or principles previously incorporated into the artificial intelligence program (Ahmed, 2022).

However, entrusting administrative decisions to artificial intelligence may lead to many errors, such as making illegal decisions, refusing to grant rights, or refusing to grant a service despite the availability of all the conditions that must be met by the person applying for this service.

In this regard, some jurisprudence believes that if the administration has taken a decision based on artificial intelligence systems as a result of an illegal decision issued by it in the first place, then it will be responsible for this decision and compensation for the harm caused to individuals (Mohammad, 2021) however when the error resulted from a technical defect in the artificial intelligence systems. In that case, the administration cannot be held accountable for this error, and it is difficult to know who is responsible for this error because it is a precise technical issue (Mohammad, 2022).

This issue has received great controversy around the world. Certain judicial systems tend to associate liability with two outcomes: 1. The administration in charge of the procedures must strike a balance between all concerned interests from the outset before the program is implemented. This calls for regular testing of algorithms (artificial intelligence), particularly if they are machine learning or deep learning algorithms. It also calls for modifications and updates. 2. The justiciability of algorithms (artificial intelligence) assumes that judges and citizens understand them, and this requires that all their aspects be known, such as their authors, preparation procedures, and decision systems (Mohammad, 2021).

Reviewing the Jordanian law in this context, we find that the Administrative Judiciary Law stipulates in Article (8) that “The lawsuit shall be filed with the court using a written summons submitted within (60) sixty days from the date of notice of the administrative decision complained of to the plaintiff or from the date of its publication in the Official Gazette or by any other means, including electronic means if the legislation stipulates that the administrative decision shall be implemented from that date.”

The researcher argues that the Jordanian legislator has specified the use of electronic notice. At the same time, he ignores clarifying the effectiveness of this notice, indicating its legal effects or the extent to which it can be considered.