It’s here! AI is poised to take over behind the scenes on crime analysis. The police are trialling software which considers features of a crime scene and then searches its database to look for similarities to other crimes. This VALCRI (a mash up of visual analytics and criminal intelligence) system then comes up with theories about how, when and even by whom, a crime may have been committed.
Of course, it’s not all plain sailing as the software must interpret interviews and other video footage as well as police reports from other cases, in which the use of language may be important. As the recent article in New Scientist highlights, a description of a suspect as “scruffy” needs to be understood and connected to “untidy” by a computer for the appropriate connections to be made.
Face recognition software is also utilised, even picking up suspects from CCTV footage. Despite assurances that ultimately “humans.. call the shots” in terms of determining the probability of a certain theory being correct, there is naturally a lot of resistance to letting VALCRI loose on new, unsolved crimes.
Professor Wong at Middlesex University, which is leading the Eu13m VALCRI project, says “It’s no longer humanly possible to sift through all the data [particularly due to the increase of social media] if the police are expected to respond in an efficient timeframe.” But it is accepted that data overload can trigger mistakes and that transparency, too, is critical in keeping VALCRI in check. Looks like Pinocchio, the software Judith and Constance encounter in The Pinocchio Brief, may have a partner in crime soon after all.