Artificial intelligence (AI) is having a moment, driven by improvements in functionality and increasing adoption by both consumers and businesses. AI is predicted to generate a tremendous amount of financial value and to disrupt business as usual across a range of industries.[1] Yet the use of AI also involves significant risks, ranging from the loss of confidential business information and personal health information to discriminatory outcomes in hiring and firing processes to physical injuries caused by autonomous vehicles. AI often involves a complex ecosystem of stakeholders, from the numerous parties spread over time and space who develop AI systems to those who integrate systems into larger platforms or physical devices to individual and enterprise end users.
When disputes involving AI arise, and these will inevitably arise,[2] it is important to be able to resolve them quickly and efficiently in a manner that results in fair and just outcomes. Alternative dispute resolution (ADR) processes such as mediation and arbitration were developed to improve upon litigation. ADR tends to be faster, simpler and more cost-effective than litigation. It also tends to be confidential, which is beneficial for reputational purposes and because it limits the risk of inadvertently disclosing trade secrets and confidential information. Parties can also select a mediator or arbitrator with particular experience and knowledge.
But ADR may be even more useful for AI-related disputes because such disputes may be relatively technically complex, expensive and involve greater risk of harmful disclosures. For example, disputes may touch on the data used to train an AI model, and that data can be enormously voluminous. The data may be a competitive advantage to a business, but sharing the data with an opposing party or expert can risk the data being publicly released. Beyond that, both parties may find it very expensive to manage information sharing, which can require the use of costly vendors and discovery software, and it may also require significant internal compilation efforts that distract personnel from their core work.
ADR can help mitigate some of these challenges, but this could be improved upon with tailored rules for AI-related disputes. The authors of this article have deep backgrounds and knowledge in both AI and dispute resolution, and are the co-creators of the JAMS Artificial Intelligence Disputes Clause, Rules and Protective Order (AI Rules). We created these rules, the first such rules released by an ADR provider, to address some of the challenges discussed above. The AI Rules have a different focus than rules related to the use of AI in the ADR process itself; for example, the use of generative AI to draft briefs or awards. In recent years, several courts and ADR providers have adopted rules or guidelines related to these concerns.
The AI Rules provide for the appointment of only panelists approved by JAMS for evaluating disputes involving technical subject matter with appropriate background and experience (Rule 15(b).) This is important because it can be particularly challenging for a generalist neutral to make sense of technological subject matter at the center of a dispute. In a best-case scenario, it may take a generalist more time to understand technical issues, which leads to additional party costs. In a worst-case scenario, not fully understanding the underlying technology can lead to a wrong legal outcome.
The AI Rules also provide built-in confidentiality and protections. (Rule 16.1(a).) They include a predetermined protective order that automatically applies (absent party agreement to the contrary), and this both helps avoid a potential dispute and removes a roadblock to early information exchange. The AI Rules also provide for more extensive confidentiality for aspects of the arbitration than most rules provide. (Rule 25(a).) This is again important to help parties avoid reputational damage from publicity associated with disputes, as well as the often scorched-earth content and tone of public filings.
Finally, the AI Rules provide a specialized process for technical experts to review materials. (Rule 16.1(b).) First, if jointly requested by the parties, the arbitrator shall designate experts who are appointed by the arbitrator to inspect AI systems or related materials, which provides critical independence and saves time and money by avoiding a battle of the experts. Expert opinions are also focused based on questions provided by the arbitrator, which avoids lengthy and expensive tangential detours. In addition, expert review of AI systems takes place in a controlled environment, which reduces the risk of information loss.
The AI Rules represent the efforts of one ADR provider to help improve dispute resolution for those making and using AI. But regardless of what rules are used, building dispute resolution into AI agreements will help parties prepare for future disputes.
Ryan Abbott, M.D., Esq., FCIArb, is a co-creator of the JAMS AI Disputes Clause and Rules and a neutral with JAMS in New York, Los Angeles and London. He has unique domestic and international expertise in the fields of life sciences, intellectual property, technology and health care. He is the author of the book The Reasonable Robot: Artificial Intelligence and the Law, which was published by Cambridge University Press in 2020.
Daniel B. Garrie, Esq. is a co-creator of the JAMS AI Disputes Clause and Rules and a neutral with JAMS with a focus on cybersecurity, data privacy, e-discovery and intellectual property. He is the founder and managing partner of Law & Forensics LLC, where he leads the cybersecurity and forensic practice teams and frequently testifies as an expert witness on e-discovery, cybersecurity and computer forensics matters. Additionally, he is a fellow of the Academy of Court-Appointed Neutrals. He is also an adjunct professor at Harvard in the School of Continuing Education, teaching Information Security, Computer Forensics and Cybersecurity Law.
