Artificial intelligence can bring benefits to legal practice, providing agility and precision. It can allow judicial decisions to be the result of the combination of algorithms, enabling the development of a system based on machine learning. This article seeks to demonstrate the current state of the use of artificial intelligence in the Brazilian justice system with the impact of the development of a deep learning system, merely the result of the automation of textual analyses of legal cases, which now serve as models. Reflection is more than necessary given the ethical issues that can arise in view of the inherent precepts that are usually impregnated in the judicial function. Civil servants, lawyers, prosecutors and judges should be guided by a pertinent regulation of new technologies and reflect on whether judicial decisions would be the result of human thinking or not, in addition to the risk that they can carry when the models are biased, in good or bad faith, due to erroneous classification or misinformation in the system.