The International Conference on Justice, focusing on current issues in the administration of justice, was held for the second time in Riyadh, the capital of Saudi Arabia, on November 23-24, 2025, with hundreds of participants representing 50 countries. This year's theme was "judicial quality". Among the 40 speakers was Zsolt Ződi, a research professor at our institute, who gave a presentation titled The perspectives and limits of using artificial intelligence in the justice systems.
In his presentation, our researcher emphasized that LLMs can be used effectively to assist judges in their work, but when it comes to complex and morally charged cases, human decision-making must always play a key role. He disputed the view of British legal futurist Richard Susskind, who argues that the "procedural approach" (which rejects the output produced by LLMs simply because it is not the result of human thought processes) should be rejected, and that we should switch to an "output approach," according to which it does not matter how the end result was produced, as long as it is useful. In human debates, the manner, course, and justification of the decision are just as important, and often more important, than the specific decision itself.