Follow us !
British and Spanish researchers have created an artificial intelligence capable of detecting lies in written complaints for theft! The program, tested for over a year by the Spanish police, is a great success! Indeed, VeriPol is effective in 83% of cases, while saving time for the police. Discouraging, its use has also reduced the number of false statements of theft.
AN ARTIFICIAL INTELLIGENCE TO HELP THE POLICE
VeriPol is an artificial intelligence, born from the research of Cardiff University in the United Kingdom and Carlos III in Madrid, Spain. The developers of this lie detector recently published their study in the journal Knowledge-Based Systems. This AI is based on a combination of automatic text analysis and deep learning, that is to say it is equipped with networks of artificial neurons that allow it to learn on its own, based on a multitude real data.
Scientists used 1122 complaints, of which only 534 were true. VeriPol has learned to recognize the typical features of false theft reports, such as lack of detail, lack of witnesses or evidence, or over-emphasis on stolen items. At present, the artificial intelligence is able to point out the lies of the complainants itself. But before presenting their results, the researchers sent VeriPol on the ground!
83% PRECISION IN REAL CONDITIONS
As of June 2017, the AI was tested in Murcia and Malaga. It was very effective in identifying 83% of false theft complaints. Indeed, in just one week, VeriPol detected 25 fake flights in Murcia and 39 in Malaga, while between 2008 and 2016, in June, the police could identify an average of only 3.33 lies in Murcia and 12.14 in Malaga.
Dr. José Camacho-Collados, co-author of the study, was delighted that the work of the teams at Cardiff and Carlos III Universities "gave a fascinating insight into how people lie to the police". He adds that "police all over Spain are now using VeriPol and integrating it into their work practices." Finally, he hopes that by showing that automatic detection is possible, it will deter people from lying to the police.