One of the important applications of artificial intelligence algorithms in the field of criminal justice is risk assessment, which is facilitated by these AI algorithms, a concept now known as "predictive justice." Predictive justice refers to a computational program capable of analyzing a set of judicial rulings and decisions to extract legal files or procedural legal texts. It also refers to AI algorithms that analyze data to extract models that accurately classify and describe key categories of data. These algorithms rely on past data to make predictions and play a significant role in helping judicial institutions achieve more reliable legal justice by linking law with mathematics, while also emphasizing that justice inherently has a human dimension.<br /><br />These AI algorithms used in such contexts are referred to as "risk assessments," and they play a crucial role in predicting the future behavior of defendants and prisoners. They can even predict the likelihood of a defendant reoffending before trial, which is known as "recidivism risk." These technologies are also used to determine the necessity and amount of bail, influence decisions related to guilt or innocence, and determine the duration of punishment, the type of correctional institution, and the eligibility for parole before the sentence ends. Risk assessment algorithms take into account various factors, such as the defendant’s age, gender, place of residence, family background, and employment status. As a result, the outcomes of the predictions may differ for two individuals who committed the same crime due to the differences in their input data.<br /><br />Legal practice in the realm of predictive justice relies on advanced computer systems that utilize powerful algorithms capable of extracting data from the internet and other diverse, verified sources of legal knowledge. These systems aim to estimate the chances of success in a lawsuit based on previous court rulings involving similar facts, regardless of the court level that issued the previous ruling.<br /><br />Countries that have adopted predictive criminal justice systems include the United Kingdom, where these algorithms have been used to detain offenders and analyze the histories of repeat offenders to prevent future crimes. This is done based on multiple indicators, such as social status and geographical location, to assess the likelihood of future criminal acts. Among the applications used is Luminance, an algorithm based on machine learning that identifies patterns, reviews documents, and learns through interaction with lawyers and legal documents. Another example is Hart, which predicts the likelihood that a suspect will commit further crimes within a specific timeframe by reviewing the suspect's history, gender, age, and geographical region.<br /><br />In France, a predictive justice program named Predictice was tested in 2017 in various civil cases at the courts of Rennes and Douai. The aim was to reduce excessive disparities in court rulings and to apply the principle of equality before the law. However, the trial yielded weak results as the program failed to provide meaningful legal insights into judicial decision-making. The program seemed to confuse factual elements with the decisive factors in judges’ reasoning.<br /><br />In the United States, such algorithms have been implemented in the field of criminal justice. For instance, Chicago developed a Strategic Subject List, an algorithm-based database analyzing individuals’ risk levels. It classified 400 individuals using a scale from 500 to 1, based on factors such as age, criminal behavior, victims, arrest records related to drugs, and gang affiliations. The data analysis showed that exposure to gun violence was a strong predictor of future criminal behavior, while gang affiliation was less predictive, and drug-related arrest history had a limited link to future crime.<br /><br />Predictive justice offers numerous advantages, including improving the justice system by reducing delays associated with traditional procedures and enhancing the quality of judicial work, both quantitatively and qualitatively. It also promotes legal awareness across various sectors, positively affecting the legal field overall. Furthermore, predictive justice supports informed decision-making by leveraging legal principles and previous judicial rulings. Over time, this contributes to greater consistency and coherence in court decisions. In this regard, Louis Larret-Chahine, one of the founders of the pilot program tested in the appellate courts of Rennes, Douai, and Lille, stated:<br /><br />"The predictive justice system will put an end to the unpredictability and randomness of justice across the country, moving towards a more logical, scientific, or at least more controlled model."<br /><br />Additional benefits include providing lawyers with clear, evidence-based insights, allowing them to offer more thoughtful, accurate, and trustworthy advice to their clients. Predictive justice can also improve the criminal justice system by enabling judges to focus their efforts on complex or novel cases, reducing the time spent on repetitive and similar disputes. Another benefit is that the system can help evaluate damages in civil and administrative liability cases, especially when compensation is difficult to determine based on geography, court type, or jurisdiction level.<br /><br />Nevertheless, predictive criminal justice is not without flaws and risks, particularly the danger of excessive reliance on it. For example, Arno de Schinisten, a lawyer at the French Council of State and the Court of Cassation, stated that the results of predictive justice applications in the appellate courts of Rennes and Douai were disappointing.<br /><br />One of the major drawbacks of predictive justice is that it reduces judicial independence, as this independence allows judges to form their convictions, see the spirit of the law, absorb its philosophy, and ultimately ensure that the will of the law prevails over personal whims and desires. This freedom is subject to oversight by the Court of Cassation, which requires judges to clearly disclose the evidence on which they base their rulings. Such evidence must be unambiguous, non-contradictory, and explicitly stated.<br /><br />As Thuraya Amrani-Makki, a professor at the University of Paris, noted:<br /><br />"Justice has not yet operated at its full potential. As long as humans, who are not infallible, program these machines, the machines too will inevitably be fallible."<br /><br />It has also been argued that predictive justice can lead to the rigidification of judicial interpretation.<br /><br />In conclusion, predictive criminal justice refers to the use of algorithms capable of modeling judicial outcomes by analyzing past decisions. It plays a role in assessing the danger posed by defendants and suspects, determining parole eligibility, and more. However, these algorithms are still surrounded by significant risks and concerns, given that they result from human programming and are prone to data bias or errors. The criminal justice system is an especially sensitive domain that involves individuals' lives, freedoms, and assets. Therefore, entrusting decisions entirely to algorithms carries substantial risk. It is more prudent for such algorithms to have a consultative role, where their input supports the judiciary in decision-making by analyzing personal and objective circumstances, such as the defendant's age, social status, criminal history, employment status, economic condition, residence, and more.<br /><br />Prof. Dr. Ammar Abbas Al-Husseini<br />Dean of the College of Law – Al-Mustaqbal University<br /><br />Al-Mustaqbal University – The Top University in Iraq