What it is: An artificial intelligence program can detect fake police reports with 80% accuracy.
How do people learn? Through experience, which means exposure to massive amounts of similar data from the same type of source. For example, police learn to detect lying by constantly exposing themselves to criminals and learning common factors that exhibit lying and telling the truth. They’re not always right but a rookie cop will be far less knowledgable than a veteran simply because the veteran has more experience. The same holds true for artificial intelligent.
Artificial intelligence (AI) learns by examining massive amounts of data to look for commonalities. Police are now using an artificial intelligence program called VeriPol that uses automatic text analysis and machine learning to identify false statements. The AI program was developed by researchers from Cardiff University and the Charles III University of Madrid, and can detect false reports with 80% accuracy.
The AI tool is able looks for false statements, such as a focus on the items stolen (usually smartphones and other high-end items) rather than the incident itself; a lack of detail which suggests that the reporter is not truly using their memory; limited details of the attacker, a lack of witnesses, and a failure to contact law enforcement immediately after. The biggest problem with analyzing police reports is that the police don’t have the time or staff to analyze fake reports so an Ain program like VeriPol can quickly identify potentially fake reports to free up personnel.
VeriPol has analyzed reports around Spain and researchers hope to test the program in other parts of the world as well. Any repetitive type of information can be detected by a trained person or an AI program. VeriPol simply demonstrates that AI can be smarter than the common criminal.