We are witnessing rapid technological progress aimed at improving the justice system’s operation. It is clear that the trend of digitizing the judiciary is growing, including tools like the common court information portal and remote hearings, which have become more popular during the COVID-19 pandemic.

In this context, Ultima Ratio is gaining interest as an online arbitration platform that focuses on resolving commercial disputes between businesses. Officially, however, the arbitration court is the Arbitration and Mediation Center at the Association of Polish Notaries in Warsaw. The platform has been in operation for about five years and has handled over a thousand cases.^[1] All proceedings are conducted completely online. The lawyer initiating the case submits a lawsuit through the platform and attaches evidence, primarily documentation.

This institution, likely a pioneer in Poland, relies on a system that utilizes artificial intelligence (AI) to generate draft justifications for arbitration awards. Although notaries present this solution as a sign of innovation and technological progress, it may, at this stage, raise certain legal, ethical, and systemic concerns.

Automation of Justifications and the Principle of Independence and Impartiality of the Arbitrator

According to information available in press materials published by Ultima Ratio and in media reports,^[1]the online system is based on an analysis of the factual circumstances of the case. It compares it with case law (both arbitration and common law) and then prepares a draft justification for the award. Consequently, the automatic generation of a draft justification for the award takes just a few seconds. Although the arbitrator has the formal ability to edit the content generated by the system, as media reports indicate, in practice, these changes tend to be minimal.^[2] This raises the question: under these circumstances, is the arbitrator genuinely performing a jurisdictional function, or is it more likely simply approving content prepared automatically?

It’s important to remember that the core of arbitration is to settle a dispute without involving a common court. Judicial authority in the case is assigned to an arbitrator chosen by the parties, who is an unbiased expert in the relevant field of the dispute. Replacing a major part of the decision-making process with an algorithm could violate this principle and, as a result, weaken the parties’ trust, especially when it is difficult to fully verify how the AI system works or the mechanisms behind its “decision-making process.”

The risk of perpetuating systemic bias and standardizing justice

Artificial intelligence systems currently learn from existing data—in this case, mainly court decisions. While this might seem like an advantage, it also introduces risks. Court rulings are not free from mistakes, inconsistencies, and systemic biases. Relying on this data, AI can reproduce the same patterns without the ability to critically analyze or correct them. Of course, the final decision still rests with the human—the arbitrator officially approves the ruling. However, the question is how much control this gives the human: has AI advanced to a point where the arbitrator’s analysis and decision become just a formality? As a result, arbitration could become a purely technocratic process of applying standard solutions, rather than resolving individual, often complex and unconventional disputes. Unlike a human lawyer, AI can invent nonexistent regulations or court rulings and lacks an understanding of the legal context.

The questionable legitimacy of the judgment and the issue of responsibility for the judgment issued.

An arbitration award is the final “product” of arbitration, representing the outcome of the arbitrators whom the parties trusted. However, a controversial argument can be made—such as for challenging the arbitrators’ decision: when the award is primarily based on an algorithm, what are the chances of challenging such a decision? This is especially relevant in cases where the justification contains errors. Who can be held responsible for these errors? Who is accountable in the context of the obligations arising from the receptum arbitrii?

Arbitrators are ultimately responsible for the awards they issue. But what if the arbitrator simply approves content prepared by AI? Are they truly accountable for such a decision? Can this responsibility be shifted to the creators of the algorithm? This question warrants careful consideration.

Protection of personal data and attorney-client privilege

Finally, it is crucial to address important issues like personal data protection, copyright, and attorney-client privilege. Remember that AI can produce content that infringes on copyright. Moreover, transferring personal data or matters covered by attorney-client privilege to AI is prohibited. When using such tools, always follow the “principle of limited trust” and carefully verify the generated content, as required in the legal profession.

Furthermore, when artificial intelligence handles sensitive or personal data, it could violate personal data protection laws, mainly the GDPR. Therefore, it’s essential to use these technologies carefully and ethically, following all relevant laws and professional standards.

Summary

The use of artificial intelligence in arbitration marks an important step in its development. The stated goals of this approach—such as increasing efficiency, reducing the length of proceedings, and lowering costs—are undeniably important. However, we must not overlook the core principles of civil litigation, including the fairness of the process. The lack of transparency, the marginalization of the arbitrator’s role, the potential for systemic errors, and the issue of accountability for the content of awards—all suggest that AI’s role in this area warrants serious scrutiny.

Arbitration, although known for its flexibility, must maintain the principles of the rule of law. Artificial intelligence can assist arbitrators, but it cannot take over their adjudicative responsibilities. Whenever arbitration becomes an institution lacking individual judgment and independent decision-making, it harms public trust in arbitration as a means of resolving civil cases.

^[1]https://ultimaratio.pl/ultima-ratio-w-mediach [access from July 8, 2025].

^[2] https://www.prawo.pl/prawnicy-sady/e-sad-polubowny-notariuszy-juz-korzysta-ze-sztucznej-inteligencji,533732.html [access from July 8, 2025]. 

^[1] Quick Arbitration for Business [access from July 14, 2025].