Artificial Intelligence can assist parties and mediators with research, document analysis and preparation, but it cannot replicate the human judgment, empathy and ability to understand underlying interests that make mediation successful, Dr. John Fletcher, Executive Director, Dispute Resolution Service, RICS, observed during a panel discussion on “The Role of AI in Mediation – Practical Use, Risk and Professional Responsibility.”
The discussion, moderated by Chris Alder, Senior Executive Officer, Standards and Regulation, RICS, featured Dr. Fletcher, Senior Advocate Priya Hingorani and Avnit Singh Arora, Director (Arbitration & Conciliation), Department of Legal Affairs, Ministry of Law and Justice, Government of India.
Opening the discussion, Priya Hingorani highlighted the limitations of pre-litigation mediation in India, particularly the absence of a consistent practice of sharing case briefs with mediators. She observed that, unlike certain other jurisdictions, mediators in India often receive little information about the nature of the dispute before the mediation begins. In the absence of adequate case material, the scope for using AI to develop a meaningful mediation strategy remains limited.
She also pointed out that pre-litigation mediation is still developing in India, with the Mediation Act, 2023 and the institutional framework around it expected to shape the process further.
Responding to Alder’s question on the potential role of AI, Avnit Singh Arora suggested examining mediation across three stages: pre-mediation, the mediation itself and post-mediation. He identified the pre-mediation stage as an area where AI could make a significant contribution.
He explained that AI could assist parties in understanding a mediator’s professional background and experience, evaluating litigation risks and possible outcomes, and developing negotiation strategies. It could also help parties assess their positions and consider alternative strategies before entering mediation. Institutions, too, could use AI in areas such as case allocation, language selection and providing parties with information about available mediators.
John Fletcher, however, emphasised the fundamental distinction between mediation and adjudicatory processes such as litigation and arbitration. He observed that mediation is essentially a negotiation between parties and not an exercise in preparing the strongest possible case before a judge or arbitrator.
According to him, parties preparing for mediation must understand not only the strengths of their positions but also their weaknesses, including the aspects on which they may be willing to compromise to secure what matters most to them. AI can help identify these aspects and assist parties in prioritising their demands and developing negotiation positions.
On the mediator’s role and the limits of AI, Priya Hingorani stressed that mediation involves human emotions, egos and confidentiality. She described AI as a “junior assistant” that could help a mediator with research, analysis and strategy, but maintained that the mediator must exercise independent judgment.
She also emphasised the importance of transparency, suggesting that parties should be made comfortable with the mediator’s use of AI. Trust and confidence in the mediator, she said, remain central to the process. AI cannot replace human contact or the relationship between the mediator and the parties.
Avnit Arora highlighted data privacy and confidentiality as important safeguards. He cautioned that mediators must understand the data-protection standards of the AI platforms they use and verify AI-generated material against their own professional judgment. He also stressed the need to make parties aware of the limitations of AI-generated outputs, particularly where a party relies on potentially inaccurate information during negotiations.
John Fletcher questioned whether disclosure of every instance of AI use should become mandatory, observing that parties are increasingly using AI in preparation for mediation just as they do in litigation and arbitration. He cautioned against imposing restrictions that could undermine the essentially consensual and confidential nature of mediation.
At the same time, he warned against uploading highly sensitive information, including material concerning minors, national security and confidential commercial matters, onto AI platforms without carefully considering the risks. He also identified the danger of inaccurate AI-generated material being presented as persuasive analysis, potentially skewing negotiations and leading a party towards a settlement it might not otherwise have accepted.
He observed that experienced mediators are particularly important in identifying such inaccuracies through their professional knowledge and the process of reality-testing. However, he cautioned against turning mediators into decision-makers or imposing on them an adjudicatory duty to verify every claim made by the parties. In his view, the process should be guided by general ethical standards and honesty in negotiations.
Priya Hingorani agreed that lawyers appearing in mediation also have an important responsibility to ensure fairness and honesty. She noted that mediators must remain alert to power imbalances, particularly in commercial disputes where one party may present AI-generated documents or contractual terms that place the other party at a disadvantage. The mediator’s experience and domain knowledge can help identify such concerns and facilitate a fairer negotiation.
Avnit Arora further distinguished between AI used for analysing documents and predictive AI used to forecast litigation outcomes. He cautioned that predictive AI could create unrealistic expectations about the likelihood of success and distort a party’s negotiating position. Where a mediator identifies reliance on a potentially hallucinated or inaccurate prediction, he suggested addressing the issue in a separate session with the concerned party rather than confronting them in the presence of the other side.
John Fletcher returned to what he described as the central purpose of mediation: looking beyond the evidence to understand the parties’ interests, needs and underlying concerns. He observed that no AI engine or algorithm could perform the mediator’s essential task of understanding what is driving the parties and identifying imaginative, creative alternatives that could help resolve the dispute.
Priya Hingorani illustrated the point with an example from a mediation involving members of a royal family. One side had relied on extensive AI-generated valuations of palaces, properties and businesses, while the other produced valuations from real estate agents, universities and chartered accountants. Despite several hearings and extensive discussions, the parties could not reconcile the competing valuations, and the mediation did not succeed. She observed that the mediator’s personal skills and ability to guide the parties remain decisive, even when AI and other research tools are available.
The panel concluded with a broader reflection on the need to use AI as an aid rather than a substitute for professional judgment, while preserving confidentiality, ethical responsibility and the human-centred nature of mediation.

