Harnessing technology: Embracing AI's potential in arbitration

As technology advances, Artificial Intelligence (AI) is rapidly transforming legal practices, including dispute resolution. In a recent interview with LegalcommunityMENA, legal experts from Charles Russell Speechlys, Thomas Snider (Head of International Arbitration), Patrick Gearon (Head of Middle East practice), Dalal Alhouti (Knowledge Development Lawyer), discussed the impact of AI on international arbitration.

They explored how AI is currently used, its potential benefits like increased efficiency and accuracy, along with potential risks that need to be considered. This insightful conversation offers valuable information for legal professionals navigating the changing landscape of arbitration with AI.

How is AI currently being utilized in international arbitration, and what specific benefits does it offer in terms of efficiency and accuracy?

AI, in the traditional sense of using computers to solve specific tasks within pre-defined rules, has been used in arbitration for many years, most notably in respect to document review and disclosure exercises. It has been common for some time now to host large data sets of potentially relevant digital information (emails, draft documents, etc.) and have a computer review the data for specific key words so as to identify the relevant material and how it relates to other information. Not only does this save time and costs, but it removes the human error that can occur with physical reviews. It can also generate relationship maps and chronologies that reveal aspects that would otherwise not be apparent.

What is different about generative AI (like ChatGPT) is the use of a computer to review and analyse data to create new documents and work product from it. Potential uses include assisting with preparing witness statements, pleadings, reports, opening and closing submissions, skeleton arguments and awards. AI can also review documents and generate cross-examination questions and strategies. It is also beginning to be used in transcription services. If used appropriately, this may have the potential to enhance the quality of arbitrations whilst saving time and money.

What are some potential risks associated with the use of generative AI in arbitration, particularly concerning the authenticity of generated documents?

As helpful as AI can be, it is open to abuse. Practitioners need to be vigilant and learn how to identify AI-generated fake images and correspondence, which are only becoming more convincing as the technology matures. Practitioners need to be aware of and know how to utilize software tools that are able to assess and verify the provenance of documents.

Using AI to generate material itself carries risks. When generating responses AI can ‘hallucinate’ details, fabricating case law and facts. This risk can be mitigated by using professional-grade AI systems, which will generate responses and give references to explain where the information deployed has come from.

Another risk is the possibility of a data breach. Uploading material relevant to a private arbitration may constitute a data breach and also breach the confidentiality of the arbitration itself. Again, care needs to be taken when choosing which AI to use and how to use it, so as to avoid such breaches.

Whilst we expect these risks to decrease over time, as AI software becomes more professional and users become more familiar with it, there is an ongoing risk that the drafting and analytical skills that junior lawyers develop when preparing documents will be diminished due to over-reliance on AI to do the bulk of the initial work. Law firms may have to identify alternative opportunities for lawyers to develop such skills.

In your opinion, how might AI, including generative AI, impact the generation of procedural orders and final awards in arbitration proceedings?

If used correctly there is no reason why arbitrators should not use AI as a tool to help generate procedural orders and even final awards. The use of precedents (i.e. templates created from documents prepared from earlier cases) is widespread and helpful, and using AI is essentially an enhanced form of this. It may result in better structured and more comprehensive awards.

Arbitrators are however ultimately responsible for the work issued in their name, and should consider notifying the parties of their intention to use AI to prevent a party arguing that the award is defective on the basis that it was written by AI rather than the arbitrator. It would also be counterproductive if the award is unnecessarily lengthy or unfocused because the arbitrator has failed to properly edit the draft prepared by AI.

Do you anticipate institutions issuing guidelines or regulations regarding the use of AI in arbitration, and if so, what might these entail?

Yes, it is likely that institutions and other industry bodies will issue guidelines and amend their rules to acknowledge the use of AI and ensure it is used fairly and appropriately. Some national courts and institutions have already issued publicly available guidance, including the Courts of New Zealand, the DIFC Courts, the Silicon Valley Arbitration & Mediation Center and the Bar Council of England & Wales.

Guidance is generally split between how to use the technology responsibly in light of its limitations and privacy issues, and the importance of using it openly and with client/parties’ consent.

With the advancement of AI technology, do you foresee human arbitrators being replaced by AI systems in the near future, or will AI primarily complement human decision-making?

No, there is no reason to believe that human arbitrators will be replaced by AI anytime soon. Even if laws were changed to allow for such awards to be enforced, the reality is that parties value being able to put their case to a tribunal, testing the case of the opposing party, and then receiving a reasoned, impartial decision at the end of the process. Using an AI tool is not a substitute for this – currently an AI tool cannot assess live evidence, balance fine legal arguments, appreciate the sources of law, or apply deep critical thinking.

Whilst generative AI may appear to be able to reason, in truth it is simply assessing the probability of what its dataset indicates ought to come next, with no appreciation of what it is actually stating or the impact of it. At the end of the day, an AI program is only as good as the data it has to work with.

Nevertheless, AI-generated awards can still potentially be of value. For example, having an AI tool assess the case merits against known precedents may help focus the parties on the strengths and weaknesses of the dispute, leading to settlement.

Given the uncertainties surrounding AI's future impact on arbitration, what steps can practitioners take to stay informed and prepared for potential changes?

Modern lawyers cannot avoid the impact of AI - even if they choose not to use it themselves, opposing lawyers, arbitrators and others will use it, and, therefore, all lawyers need to be wellversed in current AI technology, how it is used, its shortcomings and how those shortcomings may be mitigated or overcome by the use of other tools.

That being the case, lawyers should not wish to avoid AI, they should embrace it - the effective use of AI represents a potential competitive advantage, as it can enable the production of better quality work-product more efficiently. Lawyers should learn how to use prompts to generate better responses and learn the difference between the available AI tools on the market to select the most appropriate one to use. They should educate their clients about how AI is helping them work smarter
to the client’s benefit.

Dalal, as a Knowledge Development Lawyer, how do you see AI influencing the preparation of precedents and online practice tools within the legal landscape?

I believe AI has the potential to revolutionize the way precedents and online practice tools work.

Computerized precedents have been used for many years, with lawyers adding key details (such as party names and sale price) to a pre-prepared document that is then issued. However, with generative AI there is the ability to more naturally ‘discuss’ a legal issue or fact scenario with an AI (essentially providing more accurate prompts) which will not only make them easier to use but should enable them to prepare more sophisticated product in relation to a wider variety of matters.

Lawyers will still need to review the material carefully and will need to have the experience to identify errors such as missing (or unnecessary) clauses but having an additional ‘mind’ preparing a first draft and potentially spotting gaps will be invaluable, at little additional cost to the client.

Reproduced with the permission from Legal Community MENA. This exclusive interview with Suzan Taha was first published in Legal Community MENA – Harnessing Technology: Embracing AI’s Potential in Arbitration. For further information, please visit: Home – Legalcommunity MENA.