The emergence of Artificial Intelligence (AI) is changing many aspects of our lives. The judiciary is not excluded from this development. However, the use of AI today and in the foreseeable future is limited to specific tasks, whereas the work of a judge requires a broad range of different skills. Therefore, it is unlikely that the use of AI will completely replace the work of human judges from one day to another. Introducing AI assistance to the judiciary will most likely be a gradual and slow process that starts with the parallel existence of AI assistance and human judges. This article investigates the danger of hidden relinquishment of decision-making power during this phase. For this purpose, I present three different ways how AI assistance could be incorporated into the judiciary. Analysing these three scenarios, I conclude that it is separation rather than cooperation that prevents hidden relinquishment of decision-making power.