International arbitration: Is artificial intelligence the future of decision making?

Transcript

00:00:06 David Lee, Host

Hello and welcome to Podcasts by Brodies. My name's David Lee and today we're looking again at international arbitration and specifically at the future role of artificial intelligence, or AI, in international arbitration, or IA. I'll try my best not to get my AI mixed up with my IA and to help me with that goal, I'm joined by two experts in international arbitration and in dispute resolution generally. David Parratt KC and Ken MacDonald, a partner with Brodies LLP. Welcome to you both as we ask ‘Is artificial intelligence the future of decision making in international arbitration.

Ken, can we start with some definitions? Although we've covered it in a previous podcast, can you explain what do we mean by international arbitration?

00:01:02 Ken MacDonald, Partner

Well, thanks, David. International arbitration is a private means of dispute resolution by which a third-party determiner, known as an arbitrator, decides the dispute between the parties and provides a legally binding decision in the form of an award. It's been described as akin to private sector litigation. The international element comes from the fact that the disputants may be from different jurisdictions and/or the seat of the arbitration may be a different legal system from the one otherwise having jurisdiction over the parties.

00:01:42 David Lee, Host

Just coming to you, David, I just wonder if you could just give us some idea of the context in which international arbitration is most likely to be used?

00:01:53 David Parratt KC

International arbitration is normally carried out from a contractual document that the parties have entered where they have agreed contractually to resolve their differences or disputes by using the international arbitration process. That's in preference by party choice over going to court.

00:02:15 David Lee, Host

Thank you, David. Just moving from IA to AI, what do we mean when we're talking in this context about artificial intelligence, Ken?

00:02:25 Ken MacDonald, Partner

There are several definitions of AI, and these have evolved over the years as machine capability has grown exponentially. In 1950, the Turing test posited that if a machine can engage in a conversation with a human without being detected, it has demonstrated human intelligence, roll the clock forward five years to 1955, and John McCarthy suggested the term artificial intelligence in a research project proposal which described AI as a problem of making a machine behave in ways that would be called intelligent if a human were so behaving. Nowadays we might describe AI as computer systems that perform tasks commonly associated with human cognition, such as understanding natural language, recognising complex semantic patterns and generating human like outputs. If you cannot distinguish whether the AI output is from human activity or generated by machine learning, then that neatly describes where AI has got to today and is a natural evolution from the Turing test.

00:03:43 David Lee, Host

Thanks very much, Ken, very useful there. David, coming back to you, can you now talk a little bit about how AI might be used in the context of international arbitration?

00:03:56 David Parratt KC

Thanks, David. With its computing power, generative AI and what they call LLM, large language models, can be trained on a vast array of text-based data which enables the machine to learn the patterns and connections between words. Once trained, the model can generate text-based content based on parameters input by the user. So, within the international arbitration context, it could be used in searching millions of documents for key terms or assisting with the collation of information and then its presentation in any format. There are risks, we'll come to that later, for example, false data, hallucinations.

00:04:45 David Lee, Host

Thank you very much, David. So, looking a little bit more specifically, Ken, you've set the tone for what AI is David's given us some overall sort of idea of how it might come into international arbitration. What about some more specific ways in the whole process of international arbitration where AI might be used Ken?

00:05:08 Ken MacDonald, Partner

We couldn't cover in this podcast all the ways in which AI is currently being used, but there are several areas which spring immediately to mind. Arguably one of the most important enabling technologies for AI is speech recognition and through that we've got the opportunity to use AI for transcription services. So that with hearings being recorded via microphones, it’s possible to achieve real time transcript with speaker identification for all participating. In addition to transcription, AI could be used for interpretation so that the language in which the question is put to the witness can be put into the language that is agreed for the purposes of the arbitration. It's also possible to use AI for translation in document heavy arbitrations whereby a large number of documents can be converted into the language of the user or the language of the arbitration very quickly and at a fraction of the cost, if AI was not used. We can also think about using AI for legal research, it's a powerful tool and machines, of course, can analyse huge volumes of data much more quickly than any human can. Finally, I would look to the selection of arbitrator's as a potential and indeed current use of AI. So whether it's a case of the parties trying to determine who would be the best decision maker for them, or where is a panel of three, the two arbitrators, looking to decide upon who should be chair, AI can be used to research who would be the most suitable candidate for the job.

00:07:11 David Lee, Host

Thanks very much, Ken. David, do you have anything to add to that, you know, any more specifics where AI could be deployed in this area?

00:07:23 David Parratt KC

Well, just picking up from what Ken was saying about data, I could envisage it being harnessed and exploited in what's called document review i.e. going through a large mass of documents, searching for key terms or phrases or dates or persons, for example. It can indeed find that needle in the haystack and associated with that part of International Procedure, International Arbitration procedure is disclosure or documentary exchange, that Ken’s just touched on, where parties are seeking to recover documents in the hands of the other party. Now this can be one of the most expensive aspects of arbitration and dispute resolution in general and the majority of documents recovered and reviewed are not actually relevant to the issues in dispute, so a huge amount of resource is normally expended by human resource expended by going through all of these documents to try and find the key documents or the relevant documents. There has been challenge to predictive coding and computer assisted review in the English courts, in CPR, in the civil procedural context, but was dismissed. So, as far as I see it, I would think that AI can be of use in arranging material or information in an orderly manner in the pleadings or composing witness statements for review thereafter and it can already produce essay like text most people will have come across chatGPT and so in my view it's likely only a matter of time until it becomes widespread in arbitration and indeed legal drafting.

00:09:20 David Lee, Host

So, I guess what couple of things you've said there, David, it's very good at finding the needle in the haystack, it can be very speedy compared to a human. So, are you expecting to see AI use much more frequently in future in international arbitration?

00:09:41 David Parratt KC

Undoubtedly, even if it is used in the first instance to take away the heavy lifting with human secondary review thereafter. That's why I can see, at least soon it being a power to be harnessed to make human endeavour easier or cheaper for clients, but still, at least in the short term, requiring human oversight and input may be famous last words.

00:10:12 David Lee, Host

Thank you, David. Now, we'll come to regulation, which obviously is an issue which surrounds artificial intelligence in all its forms much more widely than the range of our conversation today. So, I mean, David, first, can you just tell us what regulation currently exists which is relevant when we're talking about AI in the context of international arbitration?

00:10:40 David Parratt KC

Well, the first thing to say is that regulation of this field is sparse and to a large measure, the regulators are playing catch up. There are two protocols out there, the sole protocol of video conference and international arbitration and ACAS 2020 Cyber Security Protocol for International arbitration. In August of last year, Silicon Valley Arbitration Mediation Centre published draught guidelines which are out for consultation, and in January this year in the UK, the ICO, the Information Commissioner's Office, launched its own consultation on generative AI. Lastly, in December last year, the EU passed an artificial intelligence act but which is still to be finalised and unlikely to come into force before 2026. My own Bar Council, which is my regulator as a barrister, has issued guidance for barristers using ChatGPT and other LLM’s cautioning us to always check generative outputs and not breaching data protections with inputs nor violating intellectual property rights. I thought I would give you this quote from the outgoing chairman of the bar where he said “The growth of AI tools in the legal sector is inevitable and as the guidance explains, the best placed barristers will be those who make the efforts to understand these systems so that they can be used with control and integrity. Any use of AI must be done carefully to safeguard client confidentiality and maintain trust and confidence, privacy and compliance with applicable laws”.

00:12:33 David Lee, Host

Thank you very much indeed, David. Ken, I think David's partly answered this about the question of why regulation is needed, he said now it is very sparse. So, if you can just summarise then why you think regulation will be needed in this area?

00:12:50 Ken MacDonald, Partner

Well, I think first and foremost, we need transparency and fairness in decision making. From the tribunal perspective, part of that is ensuring the provenance of submissions being made to it and ensuring that any guidance from legal sources is real and not a figment of some machine generated imagination, what we call hallucination in the AI world. Tribunals are entitled to expect proper behaviours from parties and how they present their cases, and this includes the sighting of genuine legal source. From the client perspective, privacy and confidentiality of data is paramount, and indeed, the reputation of international arbitration is at stake. If we don't get this right and protect the confidential information that's shared within the privacy of an international arbitration. Parties themselves need to consider regulation not only at the law of the seat where the arbitration is taking place, but also where they plan to enforce their reward at a later stage.

00:13:58 David Lee, Host

Thanks very much, Ken. We will move on now to looking in a simple, high-level way again at the benefits and the risks and challenges of the use of AI in international arbitration. David, you've touched on some of these already, but we'll look at the positives first. What are the main benefits that AI can specifically bring to international arbitration as you see them?

00:14:25 David Parratt KC

Well, David, big picture, it's been said by many that there must be significant cost savings from harnessing this power in AI to streamline at least the legal input into cases and to some extent, that must be right. There's also a distinct possibility, however, that the machines will perform the lawyers tasks. Richard Suskind, who's a leader in this field, last December in a Times article, highlighted not what AI can do for lawyers but what it means for the clients. He said “the main social benefit of legal AI will not be in making lawyers more efficient but empowering people who are not lawyers to handle their own affairs, I conceive AI will undoubtedly increase productivity in law firm but with a client push for quicker product and lesser cost”. So, whilst Professor Susskind can see the lawyers being replaced at some point by AI and machine learning, that certainly doesn't appear to be the case one hopes in the shorter term.

00:15:39 David Lee, Host

Thanks very much David. Ken, David has touched on probably some of the risks and challenges there already, particularly for the legal profession, but as you see them more generally again at a high level, what are the risks and challenges that AI usage in international arbitration poses?

00:15:59 Ken MacDonald, Partner

Well, David, open data uses runs the risk of confidential client information escaping into the electronic ether. So great care needs to be taken to ensure that that confidential information is managed properly and carefully to avoid data breaches, civil action for breach of confidence and indeed reputational damage. We know that the machines abhor any data vacuum and that they will make up answers very convincingly rather than not answer any question posed and this has been referred to as machine hallucination and is precisely why legal team in the arbitration need to use AI as a tool, but in a way that does not compromise human overlay in checking that the work product is in fact as it should be. There have been at least two cases in America where generative AI has been used by attorneys for their written court submissions where fictitious cases have been cited. These are cases that simply didn't exist and which the machines had convincingly held out to be true. The courts have censured those attorneys, not for using AI, which is recognised as a useful research tool, but for failing to check the machine output to ensure the provenance of the data. In that way, they fail to comply with their duties to the court.

00:17:26 David Parratt KC

Yeah, I would say there must be temptation, a bit like the attorneys, at least for some people, to let the machine come up with the answer as opposed to quality legal analysis and expertise that clients expect together with professional. So, one could perhaps envisage people who are short of time under pressure to meet a deadline with over reliance on this kind of machine learning and finding to their cost a bit like the American attorneys that the machine will invent, things, will hallucinate. I think it also raises concerns for fair processes and ethical uses, an AI product will not automatically infuse these into any products. On the contrary, of course, it's been said that maybe there's such machine learning is less subject to bias indeed, and the human equivalent.

00:18:27 David Lee, Host

Thanks very much, David. Given what we've discussed, then what are the boundaries and the limitations of AI in international arbitration? Specifically in terms of this idea of of can we have an AI arbitrator, Ken?

00:18:45 Ken MacDonald, Partner

Well, I think there's a real question over the extent to which we will accept determination by machines. It does happen, and we're told that eBay to take one example resolves 90% of their disputes using algorithms. However, an international arbitration where the stakes are a lot higher in terms of complexity and indeed value of dispute, it becomes difficult to imagine how far parties will accept an award from a non-human. It will likely be used as a tool by arbitrators, but how would an AI arbitrator control the arbitral process? Indeed, the hearing, particularly where we are dealing with aggressive or unreasonable behaviour by one or both of the parties. There are also challenges around how a panel of three arbitrators would interact if one of them was an AI arbitrator. I think the more fundamental question is whether machines can address the nuance of decision making and whether we will ever feel comfortable that justice has been done by a machine. AI lacks emotional intelligence, and the way in which the arbitration itself is conducted is also important, not just the end product in terms of the award.

It is also accepted that we don't fully understand how the machines think this is what they call the so-called black box problem and therefore there will always be some concerns about fairness and transparency in AI decision making. David, what would you add to that?

00:20:23 David Parratt KC

Well, if we can imagine an award that is either made by AI or where AI has had significant part to play in its construction, there may be technical arguments. As you know, Ken, the New York Convention 1958 is the go-to convention for reinforcement, both domestically and internationally, of arbitral awards. Even though that convention does not expressly state that arbitrators must be natural persons, there are some national arbitration laws and rules that do require tribunal members to possess certain qualities that can only be attributed to human beings, such as legal training and experience. So it throws up a question as to whether an AI award could be enforced in all jurisdictions that are subject to the New York Convention 1958. There's also, as you know Ken, a public policy exception to the enforcement of arbitral awards contained in the 1958 Convention, whereby a country can resist the national courts of a country can resist enforcement if it is contrary to the public policy of that jurisdiction. Query whether an AI award or AI Assisted award might fall into that category of being contrary to public policy and therefore reinforcement being refused. These are things that we don't yet know.

00:22:13 David Lee, Host

Thank you, David. Very interesting. Finally, both of you, David then Ken, just some final thoughts, given everything we've discussed today on what you think the future might look like for AI in international arbitration. So, David, if you get your crystal ball out first and then pass it over to Ken.

00:22:36 David Parratt KC

Well, the future as I see it with international arbitration and AI is overall is a positive one, I would say particularly from a client's point of view, this can only assist with speed and cost, and in the ultimate application may even play a major part in client driven dispute resolution itself. So I'm looking at it in a positive light. Ken, what about you?

00:23:05 Ken MacDonald, Partner

I'd agree with you, David. I think that we're going to see more capability from the machines and therefore greater potential for AI usage and whatever AI fanfares and subsequent AI winters because of the fanfares not being quite as true as promulgated by those trying to sell their wares. I think the direction of travel is only going in one way. Radical transformation awaits generally positive, but not always in ways we can predict.

00:23:39 David Lee, Host

Thank you very much to Ken MacDonald and to David Parratt, KC for your fascinating insights today on this topic, which is one we're going to be talking about a lot over the coming year.

Can AI in international arbitration increase speed and reduce costs? Probably. Can it find the needle in the haystack of documents? Yes, it probably can, but what about hallucination? What about the black box problem among the other multifarious issues that have been raised as potential problems? What does it mean for the future of lawyers? Probably the most fundamental question of all, which I think was raised by Ken, can we ever feel comfortable that justice can ever be done by a machine? So much to talk about there today, so much to think about.

This episode is part of Podcast by Brodies where some of the country's leading lawyers and special guests share their Enlightened Thinking about major issues and developments in the legal sector and what they mean for organisations, businesses and individuals across the various sectors of the economy and wider society, both in the UK and internationally.

If you'd like to hear more, you can subscribe to Podcasts by Brodies on all your favourite podcast platforms. For more information and insights, please visit brodies.com.