Soft Law, Hard Justice: Regulating Artificial Intelligence in Arbitration

Abstract

Artificial Intelligence (hereinafter "A.I.") is predicated on the notion that a computer programme can mimic all aspects of intellect and learning if they were meticulously documented. A.I. is increasingly being used in international arbitration by various stakeholders, including experts, litigants, tribunals and arbitrators at different stages. However, while A.I. has become increasingly prevalent in arbitration, there is still a regulatory vacuum that needs to be bridged. The increasing use of A.I. in arbitration to improve procedural efficiency and decision-making raises serious concerns about the lack of a strong regulatory framework.

The use of unregulated A.I. raises multiple technological challenges, which have resulted in some courts mandating the disclosure of A.I. use. These concerns are exacerbated by A.I.'s limited capacity to comprehend cultural nuances. Additionally, legal and ethical considerations arise, such as the permissible extent of A.I. usage in the arbitral process by various players, the duty to verify A.I.-generated work, the question of liability brought upon due to risks of algorithmic bias, or the leak of confidential information.

Research indicates a rise in calls for greater transparency in the use of A.I. in arbitration, and the need for regulation. While previous attempts to regulate A.I. certainly provide a starting point for the regulation of A.I. in the arbitral process, they are neither universally recognized, nor implemented across jurisdictions. In this backdrop, there is a need for an international legal instrument that either regulates, or encourages the regulation of A.I. in arbitration.

The authors in this paper argue for the regulation of A.I. in arbitration using an internationally-recognised soft-law framework. The flexibility and adaptability provided by soft law frameworks would complement the “international” and voluntary nature of arbitration. The proposed soft law framework could materialize in three ways: first, regulators could develop an international legal instrument based on which States can develop their own laws (much like the United Nations Commission on International Trade Law Model Law); second, A.I.-related arbitrations could follow the soft law guidelines as best practices, with the guidelines having persuasive value similar to the International Bar Association (IBA) Rules; or third, soft law could encourage arbitration institutions to incorporate relevant clauses on A.I. use in their procedural rules. The authors also explore the considerations which such a soft law ought to address.

The need for such a framework is not to draw boundaries, or control the innovative path of A.I., rather, it is to put in place ethical barriers and dynamic safeguards for identifiable risks, and to work in tandem with the requisite human oversight throughout the life cycles of A.I. processes. This would be a balanced approach which addresses issues of law, ethics, and technology on one hand, while encouraging creativity and flexibility in the field of arbitration on the other.

Keywords: artificial intelligence, arbitration, dispute resolution, soft law, international regulatory framework, technological challenges, accountability, human oversight, regulatory safeguards, arbitral stakeholders