As reported on the Cyber Security Intelligence website, police worldwide are using "Big Data" to predict crime and analyze criminal methods. Using artificial intelligence and machine learning, data is mined for information that can predict when and where crimes can occur. Algorithms can even predict potential perpetrators and victims.
Proponents of this technology say it provides new insights into crime when police budgets are under pressure. Critics say it reinforces over-policing of low income social or ethnic minority neighborhoods. If bias is built into the algorithm, it will yield biased results. Santa Cruz, California has banned predictive policing. Other police departments have found it does not help solve crimes. (Malcolm Gladwell's essay "The Kansas City Experiments" has insights into why predictive policing may not be helpful.)
There is also a larger issue. Is "Big Data" always good data? Large amounts of targeted data can help establish patterns and predictions, but raw data in itself ("Big Bad Voodoo Data" Bill Wilson called it in an insurance commentary) can be irrelevant and/or misleading. Artificial intelligence needs to be backed up by good, unbiased, human intelligence.