The application of advanced algorithm technologies within the sphere of justice is no longer a novelty. On the contrary, the legal approach to artificial intelligence deserves to be analyzed. As already observed by other scholars, the perception of “black box” artificial intelligence favors unconditional trust in the so-called “artificial intelligence” algorithms. This bias is further accentuated in the event that the algorithm is not entrusted to subject matter technicians (programmers, etc.) but to personnel from other fields. The analysis of man-machine interaction methods, therefore, poses the need to provide non-technical personnel with the possibility of interacting with the IT system in an easy way. This, therefore, implies two problems: the possibility of simple input and the possibility of a clear output. The analysis of the little-explored possibilities of "automatic input" and, that is, the possibility that an advanced algorithm analyzes the entire procedural documentation, is also a must. From a legal and political point of view, artificial intelligence is often identified as an excellent method to speed up work processes. This use can be assessed with an analysis relating to the state of the art in the field of AI in the ifeld of justice. Known biases related to sex, gender, and ethnicity are also analyzed. The paper concludes by identifying potential resolution methods for human biases related to the idea of a "perfect AI".