Algorithmic Arbitrators: Rethinking Legal Personality, Impartiality and Accountability in Arbitral Decision-Making

Abstract

Artificial intelligence is increasingly finding its way into legal practice, and arbitration is no exception. AI tools are already being used for legal research, document review, evidence analysis and preparation of drafts. However, a different set of legal questions arises when AI is not merely assisting an arbitrator but is expected to take the decision itself. Arbitration law places particular importance on the identity, independence, impartiality and accountability of the arbitrator. This raises the question of whether an artificial intelligence system can occupy the position of an arbitrator in the first place.

This paper examines this question in the context of the Arbitration and Conciliation Act, 1996. It considers the requirements of independence, impartiality and disclosure under Section 12, the statutory requirements relating to arbitral awards under Section 31, and the grounds for challenging an award under Section 34. It also examines the question of responsibility where an AI system produces an erroneous or biased decision. Whether such responsibility can be attributed to the developer, operator, arbitral institution or a human arbitrator supervising the system remains uncertain.

The paper further considers the position from a comparative perspective, with reference to the UNCITRAL framework, international arbitration standards and recent European developments concerning the use of AI in legal decision-making. It argues that the present Indian framework can accommodate AI as a tool to assist arbitrators, but recognising AI itself as an arbitrator would raise fundamental questions of legal status, accountability and enforceability. The paper therefore explores a model in which AI may assist the arbitral process while final responsibility remains with a human decision-maker.