Aryan Bhat and Adarsh Kumar
The influence of technology on the legal system has been profound, both in terms of the opportunities and the challenges it presents. Developments such as cryptocurrency, blockchain, smart contracts, and Internet of Things(IoT), among others, have created new possibilities which have challenged the conventional distinction of science fiction and reality, and have simultaneously raised new legal as well as ethical questions for jurists and practitioners alike. While some of these newer technologies in the legal field have merely increased efficiency and lowered costs, others actually contain the potential to introduce momentous changes in the legal jurisprudence of the concerned field.
Arbitration proceedings have been no stranger to this influence of technology, and more recently to the rise of Artificial Intelligence(“AI”). The use of AI in arbitration proceedings could be diverse, involving basic functions where it essentially plays second fiddle to the presence of human practitioners and arbitrators, by providing assistance to the arbitral process through better case management, drafting of arbitral awards, analysis and classification of evidence on record. Though functions could also be more radical in nature, such as the replacement of human arbitrators by complex AI-driven Machine Learning Systems, employing an extensive set of algorithms to decide disputes between the parties with little to no human intervention. The concept of AI Arbitrators has been a subject of much academic thought and deliberation globally, though little consensus has emerged upon the legality and expediency of the same to resolve disputes. Recently, the decision of a Mexican court in the Kleros case, has brought the idea of blockchain arbitration into focus, which underscores the growing importance as well as application of new-age technology in the field of dispute resolution. The concept of AI arbitrators is a step further than that of blockchain arbitration, as it involves a set of AI algorithms replacing, or at least sitting at par with their human counterparts in a panel of arbitrators to determine the award rather than the process being completely human-driven.
In this piece, the authors attempt to take a closer look at the provisions of the Arbitration and Conciliation Act, 1996( “ The Act”), to embark upon a legal analysis and determine whether the present law can even accommodate the idea of AI arbitrators in arbitral proceedings. Existing literature has confined the discussion to how pragmatic the involvement of AI arbitrators in an arbitral proceedings might be, such as the lack of empathy or understanding of the emotional stakes in any dispute, which may be necessary and seriously compromise with the ability of AI algorithms to arbitrate human matters. This piece, however, shall involve an analysis of whether such an appointment of AI arbitrators could even be legal under the existing Indian legal framework, setting aside the pragmatic and ethical side of the debate.
Appointment of an AI Arbitrator
At the outset, it is crucial to begin with the question of whether the Indian arbitration regime could be construed to allow the appointment of an AI arbitrator. Sections 10 and 11 of the Act provide for appointment of arbitrators. In line with the principle of party autonomy in the country’s arbitration regime, Section 10 provides freedom to the parties to appoint arbitrators of their choice, “of any nationality unless otherwise agreed by the parties.” It is pertinent to note the use of the term, “nationality” and its legal implications as against the use of a narrower concept like “citizenship.”
In State Trading Corporation v. Commercial Tax Officer, the Indian Supreme Court differentiated between the two concepts. The Court reasoned that the former term implies the jural relationship of a person, natural or artificial, with that of the State while the latter, implies the jural relationship of only a natural person with the State. So, while the former defines the civil rights of an artificial as well as natural person under international law, the latter only specifies such civil rights with respect to natural persons under municipal law. The use of the term “nationality” under the impugned provision therefore expands the possibility of an Arbitrator to being not merely a natural person but a fictitious legal entity as well. Further, the use of the term “person” to contemplate an arbitrator under the impugned provision is clearly of a wider import than a natural person only as per Section 2(42), General Clauses Act, 1897. But, the question still remains whether the ambit could be considered broad enough to include in its scope an AI, or alternatively, if an AI could be considered as a “person” under the General Clauses Act.
It is crucial to note that the aforesaid judgment of the Supreme Court was rendered with reference to companies, whose separate existence in law as an artificial person enjoying legal capacity is an admitted position. It is far more challenging to argue that an AI algorithm could be considered as an artificial person of a defined nationality, as its very legal status remains an issue undecided in itself.
Granting of legal personhood to non-human entities has historically been dictated by commercial expediency, though consensus exists that the idea of a legal personhood involves the question of whether a subject can be considered capable of possessing legal rights or duties. A prominent argument runs against the same being granted to AI much like to a corporation or a State. This is because granting a legal personhood to a corporation or a State fundamentally involves creating a fiction where a separate entity is deemed to exist through a set of documents which governs the interactions and conduct of a group of humans in an institutional setting. The subject of recognition, in itself, is not capable of any autonomy or will, thriugh which it can function without any human interaction. On the contrary, the very essence of an AI technology lies in its inherent ability to make decisions autonomously and creatively with minimal to no human intervention.
Specifically, in the instance of an AI arbitrator, the role itself requires the replacement of a natural person, so the “ agency” of the AI agency shall never succumb to that of a human being, unlike in a corporation where the “decision” of the corporation is basically the decision of a group of human beings acting in concert.
Furthermore, though an AI Arbitrator may conform to the complex set of instructions and rules fed by its human creator, and come upon a decision by applying the rules to the factual situation befor itself much like a human arbitrator, it would still suffer from the same limitations as any human arbitrator. These are the novel circumstances which are not being adequately addressed by the existing set of rules/instructions where the AI arbitrator shall have to nonetheless adjudicate and determine the disputes on the basis of extralegal considerations to arrive upon a law and pass a decision (as Hart would explain the conduct of a human judge in penumbral cases). Clearly, in such a circumstance, where the AI arbitrator arrives upon a decision, that shall be considered a decision of its own, without any human intervention or agency behind it. It shall be far from just to hold the human creator liable for the decision it would consequently arrive upon This unique facet of AI technologies is what makes it far from ideal to grant them a separate legal personhood.
Sections 10 and 11, in their present form, contemplate a non-human being as capable of appointment as arbitrator, though the same should not by itself facilitate the appointment of any AI algorithm, as an arbitrator. This is due to the lack of legal personhood granted to AI thus far, which the authors contend, should not be recognised given the special characteristics they bear, which make them capable of an agency of their own. Consequently, Artificial Intelligence is not qualified to even be a “ person” under the General Clauses Act which makes it incapable of holding any rights, duties or liabilities to serve the role of an arbitrator under the Indian Arbitration regime.
Aryan Bhat and Adarsh Kumar are Fourth Year Students at the National Law University Delhi.