ã€Global Science and Technology Report】 According to a report by the British Broadcasting Corporation on May 10, the British duron police are preparing to use the artificial intelligence system (AI) to help judge whether or not suspects should be detained. It is reported that the system divides the suspects into three levels of low, medium and high risk from the seriousness of criminal acts and has been tested by the police. The hazard assessment risk tool (Hart) data is from the records provided by the Durham police from 2008 to 2012. Some experts said that the system may be useful, but it also has the risk of mistakes, so it should carefully assess its risk. The police tested the system in 2013 and tracked the test results for the next two years, which is whether the suspect actually committed a crime. The results show that the prediction of low-risk suspects is accurate at 98%, while the accuracy of the predictions for high-risk suspects is 88%. This reflects the fact that the system’s tools are biased and more likely to classify suspects as high-risk levels to prevent mistakes and net fish. Sheena Urwin, head of the Duran Police Officer’s Criminal Justice Department, said that during the trial, the police had been monitoring Hart’s accuracy without affecting the detention decision. She said, “I think within the next two or three months, we are likely to go online to help police officers decide whether to detain or not.†Ms Owen explained that suspects with no criminal record are often not Hart is classified as a high-risk category, although if they are arrested for serious crimes, it will have a great impact on the accuracy of Hart's judgment. Professor Lawrence Sherman, director of the Cambridge Evidence-Based Medicine Center, also participated in the development of the tool. He said that Hart can be used in many situations, such as determining if it is necessary to extend detention time, whether it is possible to release the suspect before the complaint is filed, and whether they need to be detained pending trial. During the upcoming experiment, police officers will randomly select cases to use the system in order to compare the effects of using and not using the system. (Internship Compilation: Zhu Xueqing Reviewer: Liu Yang)