AI and Arbitration: Opportunities, Risks & the Way Forward from an Australian Perspective

This is the seventh and final article in Clyde & Co’s latest international arbitration series covering Artificial Intelligence (AI) in international arbitration. In this piece, Associate James Kim from our Sydney office opportunities, risks & the way forward from an Australian perspective.

Arbitration is an effective way of resolving disputes between international commercial parties. In Australia, international arbitration is regulated and promoted as a means of dispute resolution by the International Arbitration Act 1974 (Cth) (IA Act). The IA Act has been described as a pro-enforcement, streamlined arbitration regime that reflects international best practice.¹ The Federal Court of Australia, too, champions arbitration as an efficient and impartial method by which to resolve commercial disputes.² The advent of artificial intelligence (AI) is a remarkable technological development which offers much in the way of enhancing these characteristics. This article explores some of the benefits, opportunities, risks and concerns that AI presents for the future of international arbitration from an Australian perspective. 

AI as Efficiency and Dispute Minimisation

AI is progressively being utilised in a growing number of traditionally labour-intensive aspects of arbitration such as translation, document review, research, and drafting, increasing efficiency in these areas.  

Beyond that, AI can also help legal practitioners in their interactions with other parties. This is particularly true where the work product of a third party is involved. Presently, the use of real-time reporting in arbitration for transcription purposes is widespread. The process involves a stenographer writing in stenographic code at a rate of up to 280 words per minute at 98% accuracy.³ Still, many international arbitration lawyers will be intimately familiar with the disputes that can arise over differing views on the 2% that may not be accurate. These disagreements can range from minor semantic issues over an indistinct word uttered by a witness giving evidence, to major differences that influence the meaning of evidence entirely. It is conceivable that in future, a Tribunal may not need to intervene if the parties can use AI as an objective third party to resolve these differences. This could go a long way in minimising the scope of disputes and driving down costs. 

Potential uses of AI

AI also presents opportunities to drive efficiency in the court processes which can be a preliminary step in the journey to an eventual arbitration. Don Farrands KC, writing in the Handbook for Judicial Officers, published by the Judicial Commission of New South Wales, provides a useful example.⁴ Farrands KC says that one area in which AI could be used is in the context of interlocutory applications that deal with whether there has been a waiver of arbitral rights under contracts.  He goes on to explain that an algorithm could be developed which determines whether the minimum number of judicial steps have been taken by the plaintiff to conclude that there has been a waiver. The outcome of that AI decision would then be appealable to a human judge should certain criteria be satisfied or not. 

The use of AI in that way could streamline interlocutory applications and encourage parties to consider arbitration whilst providing a degree of certainty as to the accuracy of the decision. Much would, however, depend on judicial and public appetite on the use of AI as the initial decision-maker.

The Regulation of AI in Australia

Like many other jurisdictions, the use of AI in Australia is not yet governed by an overarching piece of legislation. Instead, the Australian framework is comprised of a disparate set of existing legislative instruments which combine to regulate AI, such as the Privacy Act 1988 (Cth) and the Competition and Consumer Act 2010 (Cth). 

Such regulation is consequential; whilst neither Act deals specifically with AI, the consequences of the use of AI may trigger a breach of their provisions. Speaking to this, in 2017, Rod Sims (the then Chairman of the Australian Competition and Consumer Commission) considered the hypothetical of two artificially intelligent robots engaging in sustained market collusion. Sims expressed confidence that existing laws were capable of dealing with this hypothetical situation if, for example, such collusion substantially lessened competition in breach of the law.⁵

The Artificial Intelligence: Ethics Framework, Disclosure & Judicial Perspectives

The lack of a specific piece of legislation is not to say that the use of AI in Australia suffers from a paucity of guidance. Quite apart from hard and fast legislative instruments and rules, the Australian Government’s Artificial Intelligence Ethics Framework enumerates eight core principles for the use and development of AI in Australia.⁶ They are: 

1. Human, societal and environmental wellbeing 
2. Human-centred values 
3. Fairness 
4. Privacy protection and security 
5. Reliability and safety 
6. Transparency and explainability 
7. Contestability 
8. Accountability

Of particular importance to international arbitration is the sixth principle; transparency and explainability. Pertinently, this principle includes the notion that individuals ought to be provided with responsible disclosure if they are engaging with an AI system, regardless of the level of impact. This raises interesting questions as to whether parties to an international arbitration should be required to disclose the use of AI. 

The question was recently the subject of some attention in the Supreme Court of the Australian Capital Territory. In DPP v Khan [2024] ACTSC 19, Mossop J was provided with a personal character reference for an offender during sentencing proceedings involving a criminal offence. 

Certain aspects of the reference prompted the judge to conduct a detailed analysis to determine whether the document had been written by a large language model program (such as ChatGPT).⁷ Amongst the noticeable inconsistencies, the author asserted that he had known the offender personally and professionally for an extended period despite being the offender’s brother. Mossop J opined that the use of large language models in the context of personal references for sentencing proceedings made it difficult for the court to work out what, if any, weight can be placed upon the facts and opinions set out in them.⁸ The judgment suggests that the involvement of AI may, to some decision-makers, be a factor which impacts the amount of weight given to evidence. 

The nuances of this issue are keenly relevant in the context of witness statements. In practice, a witness statement prepared for the purpose of international arbitration, being meticulously reviewed by experienced lawyers, should not contain errors as obvious as those encountered by Mossop J, even if drafted with the assistance of AI. Still, the Australian experience suggests that judges and arbitrators are, or ought to be, entitled to know that they may be engaging with the product of an AI system. Already the revised rules for the 2024 Willem C. Vis International Commercial Arbitration Moot (which garners competitors from 88 jurisdictions and is sponsored by most major arbitration bodies including the Australian Centre for International Commercial Arbitration) requires the submission of an AI disclosure statement by the parties.⁹ This signals a growing trend towards disclosure about the use of AI in arbitration. 

Conclusion

The use of AI in arbitration is an evolving and exciting field of play. From an Australian perspective, AI can help to minimise disagreements over matters such as the accuracy of transcripts and streamline and increase efficiency in preliminary arbitral matters. Still, owing to its novel nature, international arbitration lawyers must keep in mind that AI should, as a matter of best practice, be used ethically, responsibly and in accordance with the growing number of laws and regulations evolving to govern the use of AI. Whether that means that the use of AI will, at some point, be required to be disclosed in international arbitration is yet to be seen, but judicial headwinds in Australia as set out in this article may provide some guidance on the topic. 

This concludes this international arbitration series covering Artificial Intelligence (AI) in international arbitration. Our next series will cover the topic of Space Law & Arbitration. 

¹ International commercial arbitration | Attorney-General's Department (ag.gov.au) 

² https://www.fedcourt.gov.au/services/ADR/arbitration 

³ https://disputescentre.com.au/benefits-using-real-time-transcription-arbitration/ 

⁴ https://www.judcom.nsw.gov.au/publications/benchbks/judicial_officers/artificial_intelligence_and_litigation.html

⁵ https://www.accc.gov.au/about-us/media/speeches/the-accc%E2%80%99s-approach-to-colluding-robots-address 

⁶ https://www.industry.gov.au/publications/australias-artificial-intelligence-ethics-framework/australias-ai-ethics-principles 

⁷ at [39] to [42].

⁸ at [43].

⁹ https://www.vismoot.org/wp-content/uploads/2023/09/31st-Vis-Moot-Rules_FINAL.pdf