Criminal justice systems have long aimed at preventing judges’ subjectivity from having any impact on in the courtroom. A good, yet complex example is the case of sentencing, where an example of trying to minimise judges’ subjectivity are the infamous sentencing grids, used by different USA jurisdictions, which have entered the sentencing stage promising to limit judicial discretion thus eliminating judges’ subjectivity and consequently sentencing disparity. To say the promise was not quite kept is an understatement. Other systems, relying on less detailed guidelines or statutory regulation have left more room for the individualisation of sentences, but in parallel for subjectivity as well. A modern option to tackle the issue has emerged with the developments in processing big data. Big data has so far entered criminal justice at three levels: bail, sentencing, and parole. They all utilise a large amount of previously decided cases to build a strong algorithm able to predict the best possible answer to the given question in a specific case. The outcomes they offer are data-driven probabilities of requested instances. The clear answers such algorithms are able to produce are very alluring. They bring promises of a fairer system: informed decisions devoid of bias and any kind of subjectivity. There are many potential benefits of bringing together technological accuracy and human empathy: such decisions could be much more accurate and based on a sound analysis of predictive factors. Seen as more objective, such algorithms could instil the long-lost trust of the public in the fairness of the criminal justice system. Moreover, they may present an opportunity to purposefully re-shape the penal system in order to reflect progressive values and support a more humane outlook. However, there are some important considerations to be made before embarking on the big-data-saviour-ofjustice wagon which we will discuss in detail.