This blog is the second in a short series between Greenberg Traurig LLP and the London Chamber of Arbitration and Mediation (LCAM), in which we explore emerging technologies and international arbitration.
Artificial intelligence (AI) is everywhere, and it has been for some time. In truth, AI is an umbrella term and what has captured hearts, minds and headlines recently is generative AI, such as ChatGPT, which uses natural language processing. Ask it a question and it will generate a response that you might expect from a human.
For several years now, the machine learning element of AI has been deployed in law and dispute resolution. Just look at the eDiscovery tools available and the continuous active learning models deployed to assess what documents, compared to others, are more likely to be relevant to the underlying dispute. In particular, international arbitration lawyers have become accustomed to predictive coding, especially in those data heavy sectors such as construction where the number of potentially relevant documents often exceeds fifty million.
In this regard, AI is helping us tackle the issue of ever-increasing data volumes – a problem perversely created by our increasing reliance on technology. It is exciting that generative AI is unlocking massive datasets – whether they are parties’ data collections or growing datasets of judicial precedents – and allowing us to interrogate and question them in a way that feels familiar. Generative AI is giving data a voice.
However, whilst AI has the potential to open up data in a way that is quick and easy to use, it is important to understand the limitations of the AI and whether those limitations are the result of the underlying dataset, or the learning models it is using to interpret the data – the learning models having been programmed by a human.
As with those machine learning models we are already familiar with, adequately coded and trained generative AI can provide an invaluable resource to the arbitration practitioners of today and tomorrow. Beyond eDiscovery, it will likely become an integral part of legal research, arbitrator selection and precedent-based first drafts of arbitration documents.
Those arbitral bodies wanting to achieve fast, cost-effective dispute resolution for businesses will be those that ultimately facilitate and harness the use of AI, whether through its own functions or the processes it governs. But critically this must not come at the expense of the integrity of the arbitral process. This can only be achieved when the limitations of AI are properly understood, including the potential for bias and inaccuracies – a lawyer in the United States recently was reprimanded for lodging submissions written using ChatGPT, citing names of cases which neither the Judge nor opposing counsel could find, as they were entirely fictitious!
Mohammed Khamisa KC is a Shareholder, Leith Ben Ammar is Of Counsel and Johnny Shearman is a Practice Group Lawyer at Greenberg Traurig LLP.
