Artificial intelligence in dispute resolution: developments, challenges and perspectives for legal practice

Reuters

11-07-2025

July 11, 2025 - Imagine it's the year 2040. Corporate disputes are no longer resolved by lawyers. Instead, artificial intelligence (AI) records the facts from available data, automatically prepares the necessary legal documents, and sends them to the opposing party and the court. The general counsels of the companies involved in the dispute are disappointed that human judges still decide their cases, which slows down resolution and unnecessarily increases costs. However, soon AI will also be permitted to deliver judgments.
Such a future is not impossible. AI is changing the world and the way we work. Its influence on dispute resolution is constantly growing. Although we are still far from the scenario described above, using or not using AI is already significantly impacting how dispute resolution is carried out, affecting factors such as efficiency and costs.
According to the "2025 International Arbitration Survey" by Queen Mary University of London and the law firm White & Case (2025 International Arbitration Survey), 91% of respondents expect to use AI for research and data analytics within the next five years.
Does this apply to all dispute resolution professionals, including lawyers, in-house counsel, judges and arbitrators? Should AI be used for every step of the legal process? Would it be possible for companies and law firms to not use AI?
This article aims to shed light on where AI (already) adds value, the challenges, risks and pitfalls, and to provide an outlook to the future.
AI is already a catalyst for efficiency in the present day. Using AI is intended to save parties and counsel time, thereby reducing costs. These are the two biggest drivers for using AI according to the 2025 International Arbitration Survey, which is based on a questionnaire to which more than 2,400 legal practitioners, academics and interest groups from all over the world had responded.
AI can also have other positive effects, such as avoiding human error, ensuring equality of arms — that all parties have, through AI tools, the same access to legal resources — and improving the prediction of potential outcomes.
A brief overview of the potential uses of AI in dispute resolution is provided below:
•Data analysis for document production and organization, as well as document analysis.
•Legal research.
•Text editing and generation.
•Predictive analytics and litigation strategy development.
In predictive analytics, an AI is provided with court and arbitral tribunal decisions and uses the available data to forecast case outcomes and important topics. This optimizes the parties' strategy and serves as a valuable tool, for example for litigation funding, ultimately influencing decisions.
Furthermore, using AI can help avoid human errors, such as miscalculating damage amounts. AI can also level the playing field between smaller and larger entities in the legal market. Disputes that are particularly data-intensive no longer require a large number of lawyers to work on a case because AI makes data analysis easier, allowing the lawyers to focus more on the legal aspects.
The challenges arising from AI are as diverse as its potential applications and expected benefits. AI not only supports existing dispute resolution methods but also creates new problem areas.
Before addressing the challenges of using AI, one must first ask whether AI should be used at every stage of a proceeding. Very few parties to a dispute will be concerned if AI is used for process management tasks, such as setting deadlines or scheduling appointments. However, this assessment may change when evaluating evidence, such as witness statements, or analyzing legal arguments.
For example, 85% of counsel object to using AI to draft the reasoning sections of awards according to the 2025 International Arbitration Survey. Therefore, acceptance of AI currently depends entirely on its application and will continue to do so based on what AI can achieve, the extent to which the human element of a dispute remains essential, and whether AI use corresponds to the necessary due process rights of the parties involved.
As of today, there are hardly any laws or regulations in place which govern the use of AI for the decision-making of judges or arbitrators. Even if this changes, the human aspect can be expected to remain a decisive element for dispute resolution. Emotions and other human factors play a significant role in many disputes, in particular when thinking of witness testimony. It remains to be seen if and how AI can incorporate these factors into a strategy or decision.
Although it may be tempting to use AI as an absolutely objective supporter or the sole processor of a dispute, sometimes it is precisely the human factors that are decisive for resolving a dispute, whether through an amicable solution, understanding the facts, or achieving an outcome that brings legal peace to the parties involved.
Another significant concern for users of AI tools in dispute resolution is that the AI will produce erroneous results due to errors or bias that go undetected. The 2025 International Arbitration Survey found that 51% of respondents cited this as a reason for not using AI more intensively. Erroneous results produced by AI pose a significant risk to users of AI, as it may result in losing a case or a failure to fully understand their own dispute.
For lawyers and their clients, in particular, the potential liability of lawyers is likely to be a decisive factor. Who is liable if AI produces an erroneous result that goes undetected? This poses an immense cost risk for law firms and AI providers. However, this issue appears to be solvable, at least in part, if insurance is available for worst-case scenarios and if liability is clearly defined in the mandate agreement and the contract with the AI provider.
Another important issue hindering the use of AI is the risk of data breaches and loss of sensitive information for involved parties and judicial bodies. Respective data protection laws (e.g., the EU General Data Protection Regulation) and aspects of professional secrecy play a major role in this regard. However, some cases will also create new problem areas.
For example, AI tools operated by commercial or legal service providers learn and improve their results with each prompt. For obvious reasons, the service providers must assure that these learning results are not shared among different law firms using the same AI tool. Even within large global law firms, restrictions might be necessary to prevent certain users from accessing certain information.
How should a global law firm handle a situation in which an opponent in a dispute in Asia is a client in the U.S.? Even if an ethical wall exists between the people involved, if both teams use the same AI tool to ask questions about their cases and enter data, the tool will use data from both teams, thereby rendering the ethical wall meaningless.
In general, a lack of experience with and understanding of AI is an existing problem that will likely diminish in the future. Sophisticated skills of prompt engineering will become relevant to those involved in disputes. At the same time, ethical issues will arise because disputes involve people and companies entrusting sensitive data to a system whose functionality is not fully understood by the average legal practitioner.
Overall, the rapid development of AI capabilities presents a significant opportunity for dispute resolution. Whether it will ultimately lead to a "revolution" remains to be seen. The work of in-house counsel, lawyers, judges and arbitrators will change significantly in the future, though probably not for the worse. AI should allow those involved to focus on the "essential aspects" of a case and dedicate only as much time and attention as necessary to the ever-increasing amounts of data.
The challenges posed by AI will keep those involved in disputes and legislators busy for a long time. These challenges will result in numerous judgments that will further define how AI is used.
The question of whether and how the use of AI by the parties involved in a dispute should or must be disclosed also arises. For instance, should the parties to a proceeding have the option to exclude the court's use of AI? AI use can still be recognized to some extent. Consider, for example, the many instances of written court submissions citing nonexistent sources that have gone viral.
The authors see a great opportunity for the industry and are excited about future developments. However, the human element of a dispute cannot (yet?) be replaced. After all, disputes arise between people. They are ultimately decided by a weighing of arguments, which often requires consideration of soft factors, including tact and sensitivity (in German: "Fingerspitzengefühl"), in addition to hard facts. These are areas in which AI appears to have its limitations, at least in the near future.
The efficiency improvements already underway will benefit everyone involved. In this respect, at least, AI is a greater blessing than a curse.