bet365 best casino games Now，Artificial Intelligence in Medicine、It has been widely used in many fields such as business，But will the decisions made by artificial intelligence be more correct than those made by humans？Especially in the judicial field，There are research findings，In fact，The application of artificial intelligence has not improved the accuracy of judicial decisions；If the ruling is made by artificial intelligence alone，It is less accurate than a decision made by an individual。

The Harvard University newspaper "The Harvard Gazette" recently reported，In this day and age，The use of artificial intelligence based on data-driven algorithms can be said to be ubiquitous，But in many cases，Especially when the stakes are higher，bet365 live casino games People generally still make the final decision。Therefore，The key question is，Compared to decisions made independently by humans or decisions made independently by artificial intelligence，Is the decision made by a person with the assistance of artificial intelligence the best decision?。To answer this question，Researchers from Harvard University conducted a randomized controlled trial related to bail parole in Dane County, Wisconsin, USA。

The bail system occupies an important position in the American criminal justice system。The initial form of bail is to provide guarantee by others，That is, the defendant’s relatives and friends promise to ensure that the defendant appears in court on time。After entering the 19th century，With the migration of population，The connection between people is becoming increasingly loose，Using money as security (i.e. paying a deposit) to apply for bail is gradually becoming the main form。In the 1960s, there was another form of obtaining bail by signing a personal bet365 live casino games undertaking or an unsecured bond。The form through which a criminal suspect/defendant can obtain bail is generally determined by the judge。Researchers tracked bail decisions made by Dane County judges between mid-2017 and late 2019，And whether the suspect/defendant is arrested again within 24 months after being released on bail。

Research results show，There is almost no difference between the accuracy of a judge’s independent decision-making and that of a judge’s decision-making assisted by artificial intelligence；When decisions are made independently by artificial intelligence，His performance is not as good as the judge，Whether the judge makes the decision independently or with the assistance of artificial intelligence。Artificial intelligence often cannot accurately predict whether a criminal suspect/defendant will be arrested again，However, various restrictive measures related to bail will be used more severely。For example，Apply cash bail measures to more suspects/defendants，Instead of asking them to sign bet365 best casino games a personal commitment or unsecured guarantee。and，AI will recommend higher bail for non-white suspects/defendants than for white suspects/defendants。However，In more than 30% of cases，The judge did not follow the advice of the artificial intelligence。

Researchers said，This is very surprising，Because most of the existing evidence shows，The decisions made by algorithm-based artificial intelligence will be more accurate than human decisions。This may happen because the algorithm is set too strictly，Causes AI to overestimate the likelihood of suspect/defendant committing misconduct again，Believe that the suspect/defendant will frequently engage in inappropriate behavior and be arrested，Thus it is recommended to take overly severe measures。Researchers think，Artificial intelligence currently used in the criminal justice system and how humans make decisions should be studied，In order to improve artificial intelligence decision-making，And solve bet365 live casino games such problems by recalibrating the algorithm。

In the opinion of researchers，The advantage of artificial intelligence or algorithms is that they can become transparent，The key is to have open source AI that is ready for empirical evaluation and analysis。

In experiment，Researchers introduce a new methodological framework。They express the concept of "decision-maker's capabilities" as，Solving classification problems by combining a matrix of standard uncertainty and a potential outcome framework for causal inference。Decision makers themselves determine which potential outcomes should be of concern by making non-random decisions。For example，When deciding whether to collect a bond from a criminal suspect/defendant，Judges must be in public safety、Striking a balance between the administrative effectiveness of the courts and the various costs of incarceration，They need to be able to predict the possible behavior of the suspect/defendant after being released on bail based on various bet365 live casino games reference factors，And classify its social harm，Then make judgments on matters related to bail。Researchers think，This ability of judges can be defined as decision-making ability。

(Cui Xiu/compiler)