On 19 February 2025, a panel of arbitration experts gathered in New York to explore how emerging technologies—particularly artificial intelligence (“AI”)—are reshaping international arbitration. The discussion, titled “From Today to Tomorrow: Tech-Driven Arbitration,” was hosted by the AAA-ICDR and ArbTech and moderated by Mihaela Apostol (ArbTech, London).
Law Firms and AI: Adoption, Resistance, and Client Pressure
While law firms are often seen as drivers of legal innovation, adoption in arbitration remains cautious.
Marta García Bel (Freshfields, New York) offered a candid look at the realities of AI adoption in Big Law. She noted that AI-powered tools like Relativity, Disco, and Jus Mundi are already enhancing document review, case management, and legal research. But there are hurdles. “Integrating AI tools into large law firms isn’t seamless,” Marta warned. Testing new technologies takes months, and requires training programs for staff and lawyers and collaboration with tech providers to customize solutions for specific needs. In the fast-paced context of AI, keeping up with technological advancements and updates is challenging. In addition, data privacy and confidentiality concerns remain a constant challenge, particularly for clients handling sensitive information.
Interestingly, Marta noted that some clients are one step ahead of their lawyers in embracing AI. Some have even begun using generative AI tools to audit law firm invoices, flagging inefficiencies and questioning whether tasks billed by lawyers could have been performed by AI. This shift, she suggested, is pushing law firms to reconsider how they deliver and price their services.
Aníbal Sabater (Chaffetz Lindsey, New York) noted that, at his law firm, the focus has been on AI tools that assist with document production and drafting. “If you’re not already using AI for document production,” he said, “you will be soon.” He also highlighted the potential of AI in drafting pleadings and distilling case information. Particularly in investment arbitration, AI can process fact patterns and suggest case law, sometimes even draft early versions of legal arguments. However, he warned of risks, particularly the possibility of hallucinations in document drafting. Human oversight remains crucial, he stressed.
Linda Beyea (AAA-ICDR, Atlanta) noted that 82% of AAA-ICDR panelists surveyed in 2024 do not use legal AI tools, with limited uptake even among those familiar with platforms like Westlaw Precision and CoCounsel. Many are experimenting with AI tools but remain uncertain about their value.
Institutional Innovations: AAA-ICDR’s AI Tools and Initiatives
As law firms grapple with the complexities of AI integration, arbitral institutions are also innovating to enhance their services and meet evolving user expectations. Linda Beyea shared how the AAA-ICDR has embraced AI to enhance its services for both arbitrators and parties. The institution has long used technology in case management, but recent developments have focused on leveraging AI to improve efficiency and user experience.
Among these developments is the AAA’s ClauseBuilder AI tool, which uses a chat interface to assist users in drafting arbitration clauses. The tool, which has already won several legal tech awards, is powered by retrieval-augmented generation (“RAG”), a technique that enhances large language models (“LLMs”) by allowing them to access and incorporate information from external knowledge bases, improving the accuracy and relevance of their responses. Another innovation is the AAAi Panelist Search tool, which uses semantic search to enhance the list selection process. Unlike traditional keyword-based search, which matches exact terms, semantic search understands the meaning behind the query, returning results that are conceptually relevant, even if the wording differs. This reportedly reduces the time taken by case managers to prepare arbitrator lists by up to 55%. The feedback from case managers has been overwhelmingly positive. “They’ve told us the AI tools make them less biased in choosing arbitrators,” Linda said. “They’re finding people they wouldn’t normally have found.”
Linda also highlighted recent AI pilots with CoCounsel (now part of Thomson Reuters), VLex’s Vincent, and Clearbrief. Following successful trials, AAA now offers Clearbrief—an AI tool that generates hyperlinked timelines and deposition summaries—for free to arbitrators on AAA cases. Meanwhile, AAA’s intake teams have adopted Vincent to speed up contract review for new filings, reporting significant time savings.
Furthermore, AAA has just launched an AI assistant directly within WebFile, its online case management platform. The tool allows parties and arbitrators to summarize documents or query case files without leaving the system, reflecting the AAA’s broader effort to integrate AI tools seamlessly into the arbitration process.
The AAA-ICDR example demonstrates how arbitral institutions are increasingly focusing on meeting market demands by integrating AI tools into their services (see here for more examples).
Regulation and Ethical Considerations: Current Frameworks and Future Needs
The panelists turned next to the patchwork of AI regulation and ethical concerns that arbitration practitioners are navigating.
Aníbal Sabater described the United States federal landscape as largely unregulated. There is no comprehensive AI law enacted by Congress, and most recent developments have come by way of executive orders. The political deadlock in Washington has made significant legislation difficult, and there is bipartisan concern that heavy regulation could stifle AI innovation—potentially giving a competitive advantage to less regulated jurisdictions.
At the federal level, an executive order issued in 2025 by President Trump removed barriers to the use of AI in government and instructed federal agencies to develop action plans. But, as Aníbal pointed out, this remains a very general framework with few practical outcomes so far. While some states are still considering new AI-related rules, some have already enacted legislation. However, because regulatory authority in the United States is divided between the federal and state governments, state-level laws are sometimes constrained in scope, particularly when it comes to matters that affect national or cross-border markets. For now, much of the regulation comes from self-regulatory initiatives.
Turning to Europe, Sophie Nappert (3VB and ArbTech, London) explained that the EU AI Act adopts a risk-based approach. AI use in judicial and arbitral decision-making is classified as high-risk, requiring providers to meet strict transparency and quality standards. In the UK, the current government under Keir Starmer is taking a more pragmatic view, aiming to balance an innovation-friendly path with regulatory clarity. The challenge for regulators, and lawyers, is to ensure predictability without stifling progress.
The panelists agreed that existing ethical frameworks, including both legal standards and broader professional or institutional guidelines, are struggling to address the complexities of AI use in arbitration. Issues such as human oversight remain largely unresolved. Sophie raised concerns about this concept frequently invoked in the EU AI Act. In reality, she argued, it is difficult for humans to oversee AI decisions meaningfully—especially when dealing with massive datasets or complex analyses.
New Developments: Small Language Models and AI Agents
As the conversation turned towards future developments, the panel explored the emerging role of small language models (“SLMs”) and AI agents in arbitration.
Aníbal Sabater explained that, unlike LLMs, SLMs are designed for discrete tasks and trained on smaller datasets. This makes them attractive for law firms looking to streamline specific workflows, such as drafting lease agreements or generating pleadings in investment arbitration. While SLMs offer efficiency and reduce human error, for instance by minimizing routine drafting errors or inconsistencies that often arise in legal work, they require significant investment and ongoing updates. Aníbal predicted that law firms will likely collaborate with specialized providers to develop these tools, rather than building them independently.
Linda Beyea added that AAA-ICDR’s experience shows that RAG and prompting techniques, which are carefully crafted inputs that guide the model’s behavior, often deliver sufficiently accurate results. While fine-tuning a generalized model by retraining it on a specific dataset can offer greater control over its output, she noted that in practice, this approach is usually unnecessary and adds significant cost and complexity.
On AI agents, Marta García Bel described a future where AI automates tasks like document formatting, cite-checking, procedural order drafting, and document production requests, functioning almost like digital paralegals. She cited Three Crowns’ initiative to develop AI personas based on their partners’ expertise for mock cross-examinations as a leading example. However, questions remain around who bears the cost of deploying AI agents and whether their time is billable.
Rethinking the Lawyer’s Role
The panel closed by reflecting on the broader implications of AI for the legal profession. As AI tools assume tasks traditionally performed by junior lawyers and paralegals, law firms must reconsider how they train new lawyers and maintain quality control. Several panelists emphasized the importance of junior lawyers gaining hands-on experience before relying on AI, ensuring they develop the skills necessary to review and validate AI-generated work.
Sophie Nappert called for a rethink. If AI handles repetitive tasks and speeds up document production, lawyers must focus on adding strategic value, creative reasoning, nuanced argumentation, and safeguarding fairness. “We need to prompt young lawyers to think more creatively, less traditionally,” she said.
The audience discussion underscored the need for thoughtful engagement with these issues. Some raised concerns that AI could entrench conservative reasoning, limit legal innovation, diminish critical thinking, and call into question the lawyer’s value-add in an AI-driven world. Others expressed optimism that AI might free lawyers to focus on higher-value tasks and enhance efficiency.
Where Next?
Aníbal Sabater predicted that while AI is already transforming the practice of arbitration, the next major shift may not come before four or five years. By then, the panel noted, arbitration practices may look leaner, with fewer junior lawyers and a heavier reliance on AI tools.
Whether these changes lead to more efficient, accessible, and just dispute resolution remains to be seen. But as the panel made clear, the transformation has already begun.
The content of this post is intended for educational and general information. It is not intended for any promotional purposes. Kluwer Arbitration Blog, the Editorial Board, and this post’s author make no representation or warranty of any kind, express or implied, regarding the accuracy or completeness of any information in this post.
________________________
To make sure you do not miss out on regular updates from the Kluwer Arbitration Blog, please subscribe here. To submit a proposal for a blog post, please consult our Editorial Guidelines.
