Harnessing AI’s Potential For Arbitration: Prospects And Problems

While AI can augment decision-making processes and enhance efficiency, it is human arbitrators who bring to bear the wisdom, judgment, and ethical considerations necessary for equitable outcomes.

I. Introduction

In an era where artificial intelligence (AI) and big data have become pervasive, the integration of AI into our daily lives has become an intrinsic part of modern society. From driverless cars to medical diagnosis, AI is no longer just a gateway to the future; it has evolved into an indispensable component of our present reality. The legal community is already availing services such as video – conferencing, e-disclosure, legal research and contract analysis. In certain jurisdictions, AI is being employed in advanced functions such as appointment of arbitrators, estimation of costs, etc. While the development of some of these technologies promise to change the practice of law and the make the lives of practitioners easier and more productive, it remains to be seen the extent to which it will make the human role in adjudication easier or more efficient in the future.

II. Artificial Intelligence: Understanding the Fundamentals

AI is the ability of machines to learn in a manner considered intelligent by human beings. This concept is akin to the ways in which children grow and develop cognitive abilities in their early stages of development. However, unlike humans, whose learning abilities develop through a natural process of communication, machines require externally designed frameworks, i.e. algorithms, for learning. Since the frameworks designed for the machine’s learning as well as adjustments are engineered rather than innate, the intelligence acquired by them is not natural, but artificial.

Machine Learning is a subset of AI wherein the information is fed in the form of algorithms and such algorithms are used to output information. Effectively, the goal of machine learning is for the algorithm to use the information from the past (fed as data for the algorithm) to predict the future.

Deep Learning is a further subset of Machine Learning with the distinction in the way corrections take place. This design is akin to the neurons in the human brain and allows the algorithms to self-learn and resolve complex issues, similar to the neurological process in human beings.

Instances of usage of AI across jurisdictions

Interestingly, some jurisdictions have gone a step ahead and incorporate AI in taking up more substantive roles. For instance, Dispute Resolution Data, a startup in USA, acts as a worldwide repository for details regarding international commercial arbitration from various arbitral institutions such as ICC, ICDR, and the Centre for Effective Dispute Resolution (CEDR). Such details include the claim amount, location, cost, duration, and outcome of such arbitration.

Another groundbreaking usage of AI, in the form of predictive analytics, is in the prediction of legal outcomes. This involves the ability to predict case outcomes by analyzing extensive databases of precedents and filtering reported case decisions and publicly available arbitral awards on similar questions of facts and law. Examples of such software are Lex Machina and Solomonic.

Certain jurisdictions have also attempted to incorporate AI into its substantive role within its legal system as well. For instance, China has been using AI in its “internet courts” presided by non-human judges which allows participants to register and resolve their cases through online filing and digital court hearings. In Mexico AI i.e., the Expertius system assists in the determination of eligibility and valuation of pension. While these instances stand as a testament to the rapid strides taken by courts in the incorporation of AI in the justice delivery system, it remains to be seen whether this can lay down a substantial base to argue for AI to replace the human role in adjudication.

III. AI Arbitrators: Realistic or Over-ambitious?

The arguments in favour of AI arbitrators do not require a lot of emphasis or explanation. Not only does it make all our lives much easier, but it also increases the efficiency and pace of justice dispensation at a pace not fathomed earlier. The experience in China and Mexico further strengthens the case for use of AI arbitrators, at least for small-scale disputes or cases based on documentary evidence.

The use of AI arbitrators may also result in eliminating the need for extended final hearings by lawyers since AI would have the potential to streamline important arguments, filter corresponding evidence and arrange them in terms of arguments and relevancy. This would certainly result in increased efficiency and reduce the time required for an arbitration to conclude. Beyond this, the use of AI arbitrators and courts would significantly contribute towards reducing the judicial docket and increasing the access to justice.

IV. Concerns against AI Arbitrators and Way Forward

The primary concern is that factors such as induction and intuition to assess social facets are only inherent in humans and cannot be emulated by technology. It is asserted that judges do much more than to adjudicate a dispute and technology’s role to limited support decision-making, rather than supplant it. Most importantly, it puts in peril the foundational values of rule of law, which forms the basis of any justice delivery system. Such arguments are founded on the basis that AI is incapable of learning emotional values. Therefore, it is far-fetched to argue that AI, in its present state of development, can emulate the emotional and intuitive values of humans.

On the contrary, the stochastic nature of individual human judgments undertaken by human judges may be inherently less predictable than the operations of automated systems. Thus, by centralizing the decision-making process, automation may offer the potential for increased certainty rather than diminish it. This shift will not hinder but rather enhance the role of law as a facilitator of social justice.

Furthering Bias

Many critics say that AI processes can result in outcomes that are influenced by bias. They assert that technology may not only be ineffective in reducing bias where discretion is relevant, it may perpetuate such biases further. It is important to keep in mind that an AI acts as a reflection of biases inherent in the data it learns from, thereby, effectively acting as a mirror reflecting the realities of societies. The resulting biases in AI stem from the biases present in the data it is trained on. Consequently, some may argue that while bias in AI warrants attention, it is not fundamentally more problematic than the biases already prevalent in our existing systems. Having said that, it remains to be seen how such biases can be appropriately rectified without over-compensation (as seen in the recent instances of Google Gemini) and we must await such developments and undertake adequate testing before its incorporation in the legal system.

Independence and Transparency

The independence and impartiality of arbitrators forms one of the core tenets of arbitration. Since it is not clear how the training data would be prepared, such lack of clarity raises questions over its independence. This also inevitably leads to the assumption that AI typically operates in a black-box that makes the algorithm opaque to the practitioners and users and further raises the risk of awards being set aside or not enforced.

However, an AI arbitrator’s lack of emotional intelligence may actually be useful in protecting its independence. One way to address this concern would be through the incorporation of Explainable AI into the algorithm. Instead of revealing the source codes or the training data, one of the ways to address this concern would be to explain the decision-making process and reasoning steps of the AI to understand this decision.

Confidentiality & Privacy Risks

Further, another concern that has been flagged is the confidentiality and data management threats that arise with AI. To mitigate privacy risks in AI, we may consider implementing a consent-based system, wherein explicit permission is obtained from the parties involved. Additionally, AI algorithms can be designed to operate in a manner that safeguards the parties’ data from previous arbitrations, while still leveraging the rationale and legal analysis from past cases to inform current adjudications.

From a regulatory standpoint, there is a pressing need for specific recognition of AI platforms either within the framework of the Information Technology Act, 2000 or through a separate robust regulatory framework. Such measures would not only protect user data but also foster confidence in the integrity and reliability of AI arbitration systems.

V. Conclusion

The integration of AI and the arbitral community presents a landscape of promise with a need for supervision. In exploiting the scope of such promise, it is imperative to ensure that appropriate regulatory framework with procedural safeguards is robustly framed to protect the rights of the parties and ensure transparency. While AI can transform the arbitral process, any public deployment needs to proceed with abundant caution and assessment. At the heart of this discussion lies a recognition of the indispensable role of human intelligence in arbitral proceedings. While AI can augment decision-making processes and enhance efficiency, it is human arbitrators who bring to bear the wisdom, judgment, and ethical considerations necessary for equitable outcomes.

Disclaimer – The views expressed in this article are the personal views of the authors and are purely informative in nature.