THE AUTHOR:
Tiffany Lam, Communications Officer at Jus Mundi
Introduction
At this year’s Singapore Convention Week, a lively lunch session titled “AI’m Not Buying It”, co-hosted by Dentons Rodyk and Jus Mundi, put Artificial Intelligence (“AI”) in the hot seat. Amid the hype, arbitration practitioners are asking sharp, urgent questions about the practical considerations of implementing AI in their practice.
The session featured Jean-Rémi de Maistre, CEO and Co-founder of Jus Mundi, alongside Melvin See, Senior Partner & Co-Head of International Arbitration at Dentons Rodyk, and Andrew Wong, Senior Manager of Innovation at Dentons Rodyk, for a candid exchange.
The speakers tackled four key topics on the realities of using AI in arbitration practice: accuracy versus hallucinations, the costs of legal AI, data security and confidentiality, and how agentic AI is shaping the future of arbitration.
Accuracy vs. Hallucinations: Building Trust in Legal AI
Given the stakes of arbitration proceedings, reliability remains the defining challenge for practitioners adopting legal AI. The legal profession has seen no shortage of cautionary tales: citations of non-existent authorities, unverified AI outputs making their way into submissions, and the reputational damage that follows.
One of the first questions for many lawyers is, therefore, how Jus AI (Jus Mundi’s AI tool) ensures the reliability of its output and avoids hallucination. As Jean-Rémi observed, hallucination is not an unavoidable flaw of AI but a consequence of design choices. Drawing an analogy that resonated with the audience, he compared AI to nuclear energy: immensely powerful, but only as useful and safe as the plant built around it. Large language models, he explained, are like the energy source; the real question is how the engineered product channels that power into serving specific purposes.
Some AI tools are built to produce answers at all costs, even when it involves inventing information. Others, like Jus AI, prioritize precision over creativity, acknowledging its limits when sources are lacking and grounding every answer in verifiable sources. As Jean-Rémi put it, AI should reduce mistakes rather than hallucinate confidently.
Jus AI’s design leverages Jus Mundi’s extensive, structured database, relying solely on vetted sources to minimize the risk of hallucination. Every response is verifiable through line-level citations and traceable reasoning. By anchoring outputs in trusted sources, Jus AI provides practitioners with trust in the research results they use in proceedings.
AI and Costs: Investment, Access, and Value
A recurring question was whether AI would make arbitration more affordable. While law firms increasingly disclose their use of AI to clients, and in-house counsel welcome the efficiency gains, billing reductions have yet to materialize in a uniform manner.
Jean-Rémi illustrated how AI reshapes workflow rather than simply cutting hours. Just as photocopying once dominated junior lawyers’ time and then disappeared, many time-consuming tasks are now being transformed. Legal research that previously took hours can often be completed in minutes, and AI enables teams to review far more documents than would have been feasible. Rather than simply reducing overall workload, this efficiency frees lawyers to generate more ideas, explore additional arguments, and enhance the quality of their work.
Adoption trends revealed a pattern unexpected by the audience: solo practitioners and small to medium-sized firms are embracing AI faster than large firms, challenging assumptions that cutting-edge legal tech is a luxury only global teams can afford. By granting smaller teams the same research power traditionally reserved for international firms, AI functions as an equalizer in arbitration practice.
Through delivering research capacity on a scale that was previously unattainable to leaner teams, AI has the potential to democratize arbitration, reduce information asymmetries, and empower them to compete effectively on a global stage.
Data Security and Confidentiality: Non-Negotiables
Efficiency gains from AI are tempered by persistent concerns over confidentiality and data security, especially regarding sensitive client documents.
Jean-Rémi emphasized data privacy and security as non-negotiables, noting that they are the foundation of trust in legal AI. Jus Mundi enforces mandatory deletion dates for all uploaded documents, ensures that client data is never used to train AI models, and maintains full compliance with the highest international standards for security and data privacy. Every upload is ephemeral and is deleted permanently once the relevant retention date expires.
These safeguards not only meet the due diligence expectations of large law firms but also reflect Jus Mundi’s commitment to ethical AI development. This emphasis on enterprise-grade security and compliance ensures that practitioners can rely on Jus AI with confidence.
Agentic AI: A Smarter Approach to Research
The session’s highlight was Jean-Rémi unveiling Jus AI’s latest agentic version built specifically for the arbitration community. Unlike traditional AI chatbots that provide a single response, the agentic AI plans its own research steps, evaluates partial results, and decides whether to perform additional searches before finalizing its analysis.
In essence, Jus AI acts as a self-directed research agent: breaking complex questions into steps, performing parallel searches, and refining its conclusions through iteration. While the process takes minutes instead of seconds, the output is far more structured, transparent, and defensible. Future potential includes integration with case management platforms, live hearing support, and contradiction spotting in transcripts.
Adoption, Training, and Ethics: Shaping Responsible AI
From its founding in 2018, Jus Mundi’s mission has been to power global justice. Today, it partners with over 100 arbitral institutions, governments, and law firms worldwide. From treaty negotiations to providing support to arbitrators, Jus AI has been trusted by leading arbitration practitioners and international teams for their most complex cases.
Looking ahead, Jean-Rémi emphasized the importance of pedagogical and critical thinking: lawyers and students must learn not just to use AI, but to critique it, understanding inherent biases. Notably, all AI is biased. The question is how, and whether it is represented to users with transparency.
In addition to embracing transparency, Jus Mundi is committed to ethical design choices. The company has withheld certain features, such as arbitrator evaluation, until adequate safeguards are in place, reflecting a philosophy of responsibility over speed.
Conclusion: Partnership, Not Replacement
The conversation at Singapore Convention Week captured a pivotal moment: AI in arbitration is real, evolving, and transformative, but its value depends on trust, security, and specialization.
With a combination of trusted data, cutting-edge architecture, and ethical safeguards, Jus Mundi establishes itself as the credible leader in arbitration-focused AI. The goal is not mere automation, but empowerment: enabling practitioners to navigate the complexities of modern arbitration with confidence. Arbitration will not be replaced by machines, but lawyers who learn to work with AI will be better equipped to deliver justice.
About Jus Mundi
Founded in 2019 and recognized as a mission-led company, Jus Mundi is a pioneer in the legal technology industry dedicated to powering global justice through artificial intelligence. Headquartered in Paris, with additional offices in New York, London, and Singapore. Jus Mundi serves over 150,000 users from law firms, multinational corporations, governmental bodies, and academic institutions in more than 80 countries. Through its proprietary AI technology, Jus Mundi provides global legal intelligence, data-driven arbitration professional selection, and business development services.
Press Contact Helene Maïo, Senior Digital Marketing Manager, Jus Mundi – [email protected]
*The views and opinions expressed by authors are theirs and do not necessarily reflect those of their organizations, employers, or Daily Jus, Jus Mundi, or Jus Connect.
