Artificial intelligence (AI) is no longer a peripheral curiosity in arbitration. It is reshaping how contracts are drafted, disputes are managed, and awards are rendered.
This post reflects on a recent webinar hosted by Mayer Brown LLP and the Silicon Valley Arbitration and Mediation Center (SVAMC) – ‘The Next Phase of AI in U.S. and International Arbitration’ (7 October 2025). The discussion offered a compelling glimpse into where the field is heading and why arbitration may increasingly serve as the preferred forum for resolving AI-related disputes.
The self-driving contract
One of the most thought-provoking concepts discussed was the ‘self-driving contract’. Drawing on the analogy of autonomous vehicles, contracts of the future may be programmed to perform automatically under changing conditions. Parties might agree on broad objectives, for instance ‘maximise joint surplus and divide 50-50’, while an algorithm determines the specific provisions necessary to achieve that aim.
As Professor Tony Casey observed, AI’s predictive capabilities could render contracts more complete, able to anticipate and address contingencies ex ante, thereby reducing the scope for interpretation or dispute.
Yet such innovation raises questions of governance: who controls the algorithm that ‘drives’ the contract, and whose data or objective shape its logic? Here, arbitration’s procedural adaptability may prove critical, providing a forum in which algorithmic decision-making can be tested, interpreted, and validated when disagreements arise.
AI and the arbitral process
AI is also transforming the practice of arbitration itself. From document review and clause generation to research and award drafting, tools such as the AAA-ICDR’s ClauseBuilder AI and its AI Assistant in WebFile illustrate how institutions are embedding generative AI into their workflows. These systems accelerate processes, reduce cost, and enhance consistency, while leaving ultimate judgment with the human arbitrator.
The AAA’s forthcoming AI-Native Arbitrator, trained on thousands or prior awards and designed to produce draft decisions for human review, exemplifies a future of augmented adjudication, where AI’s speed meets human discernment.
As Svetlana Gitman of AAA-ICDR emphasised, aligning AI with an institution’s mission means expanding access to ADR, optimising efficiency, and shaping the next generation of dispute resolution. AI thus emerges not as a challenge to arbitral legitimacy, but as an enable of its enduring strengths.
Arbitration and AI disputes
The webinar underscored why arbitration is particularly well placed to handle AI-driven disputes. Complexity, confidentiality, and cross-border scope demand procedures that are both specialised and flexible. Arbitrators can be appointed for their subject-matter expertise; proceedings can accommodate technical evidence under protective orders (as envisaged by JAMS’ 2024 AI Rules); and sensitive data can remain outside the public domain.
Recent AAA-ICDR cases, from ‘AI-powered fraud’ claims to chatbot privacy breaches, illustrate that such disputes are already finding their way to arbitration. As AI adoption accelerates, so too will the demand for arbitrators capable of evaluating whether algorithms met contractual commitments or regulatory standards [1].
[1] Mayer Brown LLP and Silicon Valley Arbitration & Mediation Center, The Next Phase of AI in U.S. and International Arbitration (webinar, 7 October 2025), presentation by Svetlana Gitman (Division Vice President, AAA-ICDR), slide 3: “Sample Disputes We’ve Administered.”
Governance and the human in the loop
As speakers cautioned, AI in arbitration raises profound governance and ethical questions. If large-language models predict outcomes or draft awards, to what extent should human arbitrators remain ‘in the loop’? How transparent must the technology’s reasoning be? And how can the process avoid an ‘AI arms race’, in which parties deploy competing predictive tools to influence settlement dynamics?
Arbitration’s flexibility again offers an advantage. Tribunals can craft bespoke procedural orders governing AI use, for example limiting inspection of proprietary systems to independent expert under secure conditions (as envisaged in JAMS Rule 16(1)(b)). In doing so, arbitrators can ensure that innovation proceeds within an ethical, balanced, and controllable framework.
Conclusion
The next phase of AI in arbitration will be defined less by technology than by trust. As arbitral institutions experiment with blockchain authentication and AI-driven drafting tools, transparency, accountability, and human oversight must stay at the core. Arbitration’s ability to evolve while preserving procedural integrity, positions it to shape global standards for AI justice.
For practitioners and stakeholders alike, the message is clear: understanding AI’s implication is no longer optional. Whether advising on algorithmic contracts, sitting as arbitrators in AI-related disputes, or deploying AI to streamline proceedings, professionals must adapt swiftly. The fusion of human judgment and machine intelligence will redefine arbitral decision-making.