AI in arbitration: Enhancing efficiency, safeguarding integrity

Generative artificial intelligence, or gen AI, is transforming the future of work across several sectors, with the legal field being a prominent one among them and, in particular, arbitration. Gen AI has the potential to make arbitration a more efficient and cost-effective process and is already being utilised by both counsel and arbitrators for many applications.

In this article, we consider some of the benefits, challenges and potential solutions.

In our field, we are mainly seeing generative AI being used for legal research, drafting simple documents, conducting document reviews, summarizing large amounts of information (including for the purpose of preparing chronologies), as well as for assistance with drafting emails and other procedural aspects.

As technology evolves further, gen AI may be utilised to summarise large sets of pleadings, generate case theories, generate strategies for cross-examination based on witness statements or expert reports, and assist arbitrators in generating procedural documents, which at the moment tends to be done by the tribunal secretary.

Such applications will continue to boost efficiency and free up human capital in the legal sector for more knowledge-based and value-added work. However, the use of gen AI in arbitration brings to the forefront several important considerations and limitations.

Accuracy of outputs from AI

With the incorporation of AI in legal processes, it becomes vital to ensure the accuracy and reliability of the outputs it generates, especially in complex arbitration cases.

The fundamental principles underpinning current gen AI technology rely heavily on the input that it is fed being accurate and reliable. In a sense, it follows the phrase, “Garbage in, garbage out.”

Therefore, until the inputs that are being fed to the system have been tested for accuracy and reliability, it would be a grave error for parties to rely exclusively on the content generated by gen AI without verification by a human being.

Data privacy

Another key area to address is data privacy and confidentiality, especially considering the sensitive nature of information handled by legal practitioners.

Many publicly available, free-to-use gen AI systems have servers located in various countries, and these systems typically share no liability for any breach of such data that may occur in those countries. This reality, combined with the fact that many personal data laws restrict the transfer of data outside a jurisdiction without consent, poses limitations to the responsible use of gen AI systems.

This is particularly true in the arbitration field, the hallmark of which is privacy and confidentiality. Parties must be careful to ensure that the use of AI systems does not inadvertently lead to sensitive information being disclosed to anyone outside the arbitral process.

Transparency and regulations

Underpinning the issues of accuracy, privacy and confidentiality is transparency when it comes to the use of AI in arbitration.

Law firms should discuss the use of gen AI with clients to ensure the latter are aware of the incorporation of such systems and have any questions answered upfront. While many clients would welcome the use of innovative technologies to save time and costs, they would also expect that the legal fees they pay are not solely for the answers generated by a computer system, but for the lawyers’ strategic advice on the case.

As for arbitrators, it is even more critical that any use of gen AI in the decision-making process is disclosed upfront to the parties and only implemented with their consent. This is because one of the pillars of arbitration is party autonomy in deciding how the arbitration is conducted.

Transparency is especially important for arbitrators who may use gen AI for aspects of producing arbitral awards (decisions).

As a recent case in the UK courts has demonstrated, the rise of fake arbitral awards is an increasing threat to the integrity of the arbitral process.

Therefore, we anticipate further guidelines and regulations being issued to regulate the usage of gen AI in the award writing process. In fact, many arbitral institutions and professional bodies have already issued guidelines on the use of gen AI in arbitration for parties to then adapt for their particular situation – rather than outright prohibiting the use of gen AI in the field. This approach retains flexibility for parties to decide how best to deploy AI, tying back to the importance of autonomy.

Closing thoughts

The use of AI will become the norm in certain aspects of the arbitral process, such as the initial review of evidence, producing first drafts of documents and potentially the creation of procedural documents.

However, there needs to be a concerted effort by the users of arbitration to ensure that the use of gen AI in the decision-making process is regulated and transparent so that arbitrators remain accountable for the awards they issue.

Otherwise, the primary role of the arbitrator in issuing a well-reasoned arbitral decision based on their analysis of the facts and experience of similar issues might come under threat as parties simply rely on AI without scrutiny.

The writer is a partner at Addleshaw Goddard.