AI as a mechanism for Dispute Resolution

Question: What are the Implications of Artificial Intelligence on the Dispute Resolution mechanism of International Arbitration

Introduction

Background

According to the Silicon Valley Arbitration & Mediation Center (SV AMC), Artificial Intelligence (AI) is defined as any computer system that “performs tasks commonly associated with human cognition, such as understanding natural language, recognizing complex semantic patterns, and generating human-like outputs.” In simpler terms, AI is a computer system miming human cognitive abilities. Integrating AI into international arbitration has garnered increasing attention as AI technologies advance and reshape various industries, including the legal field.

International arbitration is a private, non-national dispute resolution mechanism that allows parties to seek fair resolutions to their disputes through an impartial tribunal. AI’s potential benefits in this context include streamlining processes, reducing costs, and enhancing decision-making accuracy. However, the involvement of AI in arbitration also poses challenges that must be addressed to maintain the integrity of the process.

Purpose of the Essay

This essay will explore the opportunities and risks associated with AI in international arbitration, examining its current applications, potential future uses, and the regulatory measures that could govern its implementation. The primary argument of this essay is that the advancement of AI in the legal field, particularly in international arbitration, should be encouraged due to its potential for improved efficiency, better management, enhanced arbitrator selection, and effective drafting.

The essay will also provide examples of AI’s capabilities such as document management and review, drafting awards, arbitration selection and management of proceedings.

In contrast, this essay will address the risks and challenges posed by AI, including issues related to bias, inaccuracy, due process, integrity, and privacy, along with examples of challenges that have already emerged in this area.

Opportunities of AI in International Arbitration

AI’s application in international arbitration is primarily seen in areas such as document review, arbitrator selection, case management, and even drafting arbitral awards. These applications promise to enhance the efficiency of arbitration proceedings and reduce the significant time and costs often associated with traditional methods.

Enhanced Efficiency in Document Management and Review

AI tools offer enhanced efficiency by automating document management and review which is a time-consuming task when done by humans in arbitration. A recent study by Goldman Sachs has revealed that 25% of all work-related tasks could be automated through AI and this figure rises to 44% when it pertains to the legal domain. This reality can be seen in law firms such as Allen & Overy using Harvey AI which is trained in legal data, including case law and reference material to complete tasks like legal research, contract analysis, due diligence or litigation. What this shows us, is that regarding AI and document management and review, AI tools can quickly analyse large batches of data, identify relevant documents, and categorise them for easy access.

This reduces the time arbitrators, counsel, and parties must spend on sifting through irrelevant or redundant materials, thus speeding up the entire arbitration process. Therefore, if the process is more efficient, the integrity of the process should also be preserved.

Arbitrator Selection

An emerging process that AI is revolutionising is the synthesis of vast amounts of data related to past decisions, expertise and tendencies of potential arbitrators to be able to select the right arbitrator for the dispute for a fair and successful arbitration to result. On the 30th of April 2024, the Silicon Valley Arbitration & Mediation Center published the final Guidelines on the Use of Artificial Intelligence in Arbitration and guideline 6 states that “An arbitrator shall not delegate any part of their mandate to any AI tool. This principle shall particularly apply to the arbitrator’s decision-making process. The use of AI tools by arbitrators shall not replace their independent analysis of the facts, the law, and the evidence.” which is important to “preserving the human element essential to the fairness and integrity of the arbitration.” What this indicates is that according to these guidelines, it would be within their scope for AI to be involved with arbitrator selection, but not with the decision-making process discussed in the dispute. Moreover, if arbitrators were selected according to AI, it could provide a more objective basis for selecting arbitrators, reducing the potential for bias and increasing the diversity of arbitrators appointed, which enhances the legitimacy of the process, as it can historical rulings and other relevant data. Therefore, preserving the integrity of the arbitral process.

Management of arbitration proceedings

AI has the potential to enhance procedural efficiency by streamlining internal processes. This is currently being introduced by the International Chamber of Commerce and the American Arbitration Association by automating administrative tasks, optimising timelines and improving communication between parties. As AI continues to be introduced within the legal field, guideline 3 proposed by SV AMC focuses on disclosure. According to the guidelines “decisions regarding disclosure of the use of AI tools shall be made on a case-by-case basis taking account of the relevant circumstances, including due process and any applicable privilege” Potential disclosure details include: “the name, version, and relevant settings of the tool used, a short description of how the tool was used and the complete prompt (including any template, additional context, and conversation thread) and associated output”. The automation of these internal processes that take place within an arbitration could reduce costs for participants and provide faster dispute resolution for parties as the most time-consuming tasks will be completed by AI. Therefore, the disclosure process would allow for there to be some oversight by the arbitral tribunals to ensure full transparency within the arbitral process. Therefore, preserving the integrity of the arbitral award.

Drafting Arbitral Awards

As the use of AI is being explored within the arbitral field, AI is being used to assist in drafting arbitral awards. This is being tested in several jurisdictions via judges using AI tools to support the drafting of judicial decisions, likewise, AI can aid arbitrators in generating drafts of awards by analysing past decisions and applying relevant legal principles. In response to the introduction of AI in arbitration, the Organisation for the Harmonisation of Business Law in Africa has been interpreting Article 5 of the Uniform Arbitration Act (UAA) to expand its scope to include AI tools. Article 5 states that the ‘duties of an arbitrator may only be performed by a natural person’ meaning that while AI cannot replace the human judgment required in the final decision-making process, it can reduce the time required to prepare decisions, thus improving the overall efficiency of arbitration. Furthermore, UAA can be interpreted as affirming that the arbitrator itself cannot be an AI programme but it can assist in the work performed by an arbitrator such as drafting arbitral awards; which can impact the validity of that arbitral award as a means of preserving the integrity of the arbitral award.

Risks and Challenges of AI in International Arbitration

Despite the significant opportunities AI presents, its use in international arbitration is fraught with risks that could undermine the fairness and integrity of proceedings. These risks need to be carefully managed to avoid adverse consequences for arbitration users in order for the integrity of the arbitral process to persist.

Bias and Data Limitations

The use of AI tools in the arbitral field invites a risk of biases to be replicated or amplified leading to unfair outcomes. Since AI tools are developed via training of a computer system, this training is conducted by humans and thus it is important to implement safeguards to potential biases of their creators and the underlying data set on which they have been trained. These biases can be seen via AI tools such as LexMachina which offers insights into extensive judgement databases to refine case assessment by promoting the analysis of courts, judges, opposing counsel and parties and generates complaint summaries and anticipates behaviours and outcomes that different legal strategies might produce. What AI tools such as LexMachina offer is efficient yet the risks of potential biases can result in replicated or amplified biases which can lead to unfair outcomes. Furthermore, since commercial awards are not publicly available, the data on which AI tools rely may be incomplete thus offering a limited scope and accuracy of analysis.

Although these risks exist, the use of AI within the legal field has capabilities beyond our imagination. Therefore for AI to persist as a helpful and useful tool within the legal field, bias and data limitations will continue to be a challenge that those programmers will have to keep working on for the integrity of the arbitral process to persist.

Hallucinations and Inaccurate Outputs

A risk of generative AI is a phenomenon called “hallucinations’ which could lead to the introduction of fabricated information or incorrect legal analysis during arbitration. Hallucinations are AI-produced responses or outputs that are not grounded in reality or the provided source material. In the New York case of Mata v. Avianca, Inc., 678 F . Supp. 3d 443, Mata’s lawyer Steven Schwartz was found to have used ChatGPT to help in his legal research while representing Mata, while stating their brief in court, Schwartz was found to have citedChatGPT generated case law that turned out not to exist, hence a hallucination. What the matter of Steven Schwartz and many similar incidents that have happened since is that the potential for AI-generated outputs to become untethered from accurate data represents a major concern, especially if such outputs influence the decisions of arbitrators or parties involved in the process. Therefore, it is fundamental that hallucinations and accurate outputs be controlled and approached with caution for the integrity of the arbitral process.

Privacy and Confidentiality Concerns

Arbitration proceedings often involve sensitive and confidential information and with the emergence of AI tools that can analyse, synthesise and review big amounts of information and data, these publicly available models raise privacy concerns as they store or access personal data entered by users. To mitigate this risk, guideline 2 of the SV AMC regarding ‘safeguarding confidentiality’ states that ‘all participants in international arbitration are responsible for ensuring their use of AI tools is consistent with their obligations to safeguard confidential information (including privileged, private, secret, or otherwise protected data)’. Therefore, in the legal profession where confidentiality is paramount, the use of AI tools could expose parties to risks of data breaches or unauthorised sharing of confidential materials. This means that legal professionals must exercise caution when submitting confidential data to generative AI tools like ChatGPT (which have been known to collect data provided by users), and should aim to use AI tools specifically designed for legal practice such as Harvey and Lexis+AI which employs safeguards like data encryption to protect client information and due process. Therefore, privacy and confidentiality are essential to preserve the integrity of the arbitral process.

Integrity of Evidence and Manipulation

Another risk of AI technologies is the use of deepfakes which introduces the risk of manipulated or false evidence submitted in arbitration proceedings. Deep Fakes are artificially generated video or audio which could be used to falsify evidence, which undermines the integrity of the arbitral process. In Dadras International v Iran Case Nos. 213 And 215, a precedent of a heightened standard of proof of “clear and convincing evidence”, what this means is that the evidence that can be presented before an arbitral tribunal must be clear and convincing to not allege forgery and cause implications of fraudulent conduct and intent to deceive. Therefore, when it comes to AI tools, they do make it easier to manipulate digital evidence, which could lead to disputes regarding the authenticity of evidence presented in arbitration and other legal proceedings. Furthermore, with the introduction of regulatory legislation such as the European Union (EU) AI Act, the Act aims to regulate high-risk AI systems and impose requirements for general-purpose AI tools. AI systems used to create deepfakes will be subject to the rules andregulations within the EU AI Act as a way for the Act to ensure compliance with fundamental rights and safeguard privacy regarding international arbitration. Therefore the protection of evidence and discouragement of manipulation is essential to preserve the integrity of the arbitral process.

Conclusion

This essay has explored how AI is rapidly becoming an integral part of international arbitration and offers substantial benefits in terms of efficiency, cost reduction and decision-making accuracy. In the first section, this essay explored the enhanced efficiency in document management and review which is a practice that is already taking place within the legal profession as AI can quickly analyse large batches of data and identify relevant documents. Next, this essay explored how AI can aid in arbitrator selection via its ability to process relevant data and historical rulings offering a more objective basis for selecting arbitrators, reducing the potential for bias and increasing the diversity of arbitrators appointed. Thirdly, this essay explored the management of arbitral proceedings which AI can enhance procedural efficiency by streamlining internal processes by completing time-consuming tasks such as administrative tasks, optimising timelines and improving communication between parties. Lastly, this section explored how AI can aid arbitrators in generating drafts of awards by analysing past decisionsmand applying relevant legal principles which according to up-and-coming regulations may question the integrity of the arbitral process and the validity of the arbitral award.

In the second section, this essay explored the risks and challenges of AI in international arbitration which could lead to undermining the fairness and integrity of arbitration proceedings. Firstly, this section analysed the bias and data limitations of AI which if not properly trained and surveilled for ethics can replicate or amplify biases, leading to unfair outcomes. Next, this section analysed hallucinations and inaccurate outcomes which can produce responses or outputs that are not grounded in reality or the provided source material which can risk undermining the integrity of the arbitral proceedings as incorrect information can be cited if not manually vetted. Thirdly, this section explored privacy and confidentiality concerns of AI in international arbitration which when regarding the legal field is imperative, meaning that legal professionals should exercise caution when submitting confidential data to generative AI tools so as not to risk exposing parties to data breaches or unauthorised sharing of confidential information and ensure the integrity of the proceedings and due process. Lastly, this section explored the integrity of evidence and manipulation focussing on the existence of deepfakes which if not properly regulated and complied with, risks undermining the integrity of the arbitral process.

In amalgamation, as AI continues to evolve, its role in arbitration will undoubtedly expand, but the arbitration community must approach its adoption cautiously. By implementing safeguards, encouraging transparency, and developing appropriate regulatory frameworks, AI can be harnessed responsibly to improve and adapt arbitration while preserving the integrity of the dispute resolution process.

Bibliography

Primary Sources

Cases

Dadras International and Per-Am Construction Corporation v. The Islamic Republic of Iran and Tehran Redevelopment Company, IUSCT Case Nos. 213 And 215

Mata v. Avianca, Inc., 678 F . Supp. 3d 443

Table of Legislation

Silicon Valley Arbitration & Mediation Center Guidelines on the Use of Artificial Intelligence in Arbitration (1st Edition)

Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024

Secondary Sources

Blogs

Haesler J and Isler T, ‘Navigating the Main Impacts of Artificial Intelligence in International Arbitration: Insights from the ICC YAAF Workshop’ (Kluwer Arbitration Blog, 17 March 2024)
https://arbitrationblog.kluwerarbitration.com/2024/03/17/navigating-the-main-impacts-of-artificial-intelligence-in-international-arbitration-insights-from-the-icc-yaaf-workshop/ accessed 22 November 2024.

Journal Articles

Evans DL, Guillon S, Losey R, Washington V , and Yancey LG, Dispute Resolution Enhanced:
How Arbitrators and Mediators Can Harness Generative AI (American Arbitration Association, 20 February 2024).

Habib S, ‘The Use of Artificial Intelligence in Arbitration in Africa – Inevitable or Unachievable?’ (International Bar Association, 2021)

Reports

Nairobi Centre For International Arbitration, ‘Artificial Intelligence “AI” in International Arbitration: Machine Arbitration’ (August, 2021).

Online Sources

Calthrop A and Limond K, ‘Artificial Intelligence in Arbitration: Evidentiary Issues and Prospects’ (A&O Shearman, 12 January 2024)
https://www.aoshearman.com/en/insights/artificial-intelligence-in-arbitration-evidentiary-issues-and-prospects accessed 22 November 2024.

Dr. Markus Altenkirch RH, ‘The New Guidelines on the Use of Artificial Intelligence in Arbitration: Background and Essential Aspects’ (Global Arbitration News, 4 November 2024)
https://www.globalarbitrationnews.com/2024/05/15/the-new-guidelines-on-the-use-of-artificial-intelligence-in-arbitration-background-and-essential-aspects/ accessed 22 November 2024.

Friedman E and others, ‘Generative AI: Opportunities and Risks in Arbitration’ (Freshfields, 2024)
https://www.freshfields.com/en-gb/our-thinking/campaigns/international-arbitration-in-2024/generative-ai-opportunities-and-risks-in-arbitration/ accessed 22 November 2024.

Garrido CR, ‘Artificial Intelligence and International Arbitration: Uses and Challenges’ (Uría Menéndez, January 2023)
https://www.uria.com/en/publicaciones/8501-artificial-intelligence-and-international-arbitration-uses-and-challenge accessed 22 November 2024.

Gopalan S and Bhagnani P, ‘Jams Unveils New Arbitration Rules for Artificial Intelligence Disputes’ (White & Case LLP, 3 May 2024)
https://www.whitecase.com/insight-our-thinking/jams-unveils-new-arbitration-rules-artificial-intelligence-disputes accessed 22 November 2024.