Stevenson, Wong & Co. Co-Hosts Pivotal Panel on AI in International Arbitration at Seoul ADR Festival 2024

Panellists and Discussion Highlights

The session featured panellists from firms across Asia and North America, including Elizabeth Chan of Stevenson, Wong & Co., Grant Kim of LimNexus LLP, Sunyoung Kim of Baker McKenzie, David McArthur of Yulchon, and Lexi Takamatsu of Mori Hamada & Matsumoto.

They engaged the audience on some of the most pressing issues involving AI-assisted work product in arbitration, examining AI’s potential to assist in the arbitration process while highlighting the essential safeguards needed to maintain fairness, transparency, and procedural integrity. The panel also discussed the practical application of the SVAMC Guidelines on Using AI in Arbitration (SVAMC Guidelines), which are the only comprehensive set of guidelines on using AI in arbitration generally as at the time of this article.

AI Assistance in Drafting Arbitral Awards

Opening with a discussion on AI’s role in assisting with arbitral awards, the panel emphasised that while AI can enhance efficiency, human tribunal members should not delegate decision-making authority to AI tools. Under the SVAMC Guidelines, arbitrators should ensure that AI complements rather than replaces their independent judgement.

The panellists debated the merits of transparency, noting that although the guidelines do not mandate a universal duty to disclose AI use to participants in an arbitration proceeding, arbitrators might consider proactive disclosure in certain cases to maintain trust in the arbitration process.

A parallel was drawn between controversial issues surrounding human tribunal secretaries’ assistance in drafting awards and similar concerns raised by generative AI tools. For example, panellists explored whether it could constitute improper delegation if an arbitrator used an AI tool to identify the top three pieces of evidence from a hearing transcript on a particular issue, even while maintaining overall authorship of the award.

AI Assistance in Expert Work Product

The panel then addressed the use of AI in preparing expert reports. Discussion centred on the role of cross-examination as an additional safeguard, with panellists noting that if an expert disclosed AI assistance in their report, it could prompt tactical questioning about the AI’s specific use and the extent of human verification of the work product.

AI in Document Production

The conversation moved to AI’s application in document production, with panellists noting that several existing tools, such as Relativity and Sightline, are widely used in e-discovery, and newer tools like Lawdify or Co-Counsel are being developed specifically for legal use.

The panel explored potential challenges if a party questions the completeness or sufficiency of a document production where AI tools were used.

While complete accuracy in document production is rarely achievable, panellists noted the risk of guerrilla tactics, where parties might make tactical challenges based on allegations of algorithmic bias and other issues.

The panel emphasised the need for arbitrators to have the technical competence to manage these situations effectively, ensuring both efficiency and the legitimacy of the process.

Confidentiality and Data Security

Confidentiality and data security emerged as central concerns, especially given AI’s potential use to handle vast amounts of sensitive information in arbitration. The SVAMC Guidelines underscore that all participants must ensure AI tools meet appropriate confidentiality requirements, favouring privacy-appropriate solutions whenever possible.

The panellists discussed practical actions to secure confidential information, including anonymising data, using AI tools with robust security protocols, and avoiding tools that cannot guarantee confidentiality.

This approach aligns with the SVAMC Guidelines’ broader emphasis on data protection, offering parties a framework for addressing security concerns in the event of a data breach. Where confidentiality lapses occur, steps such as audits or remedial assessments of AI tools may be warranted.

Conclusion

In closing, the panel reinforced the SVAMC Guidelines as a progressive framework for addressing the complex and evolving risks associated with AI in arbitration. These Guidelines represent a vital resource for the arbitration community, enabling practitioners to harness AI’s benefits responsibly while preserving core principles of fairness and procedural integrity.

Stevenson, Wong & Co. was delighted to be part of the 2024 Seoul ADR Festival, especially given the close trading ties between Korea and Hong Kong.

If you have any questions about this event or our firm’s arbitration practice, please contact Partner Heidi Chui.