The online polygraph has learned to identify liars by the way they speak
Artificial intelligence has learned to distinguish truth from falsehood. On the basis of Florida State University, an online lie detector based on a neural network has been created, which is an alternative to existing polygraphs. The device can detect lies by monitoring the words that the user selects and the speed at which he types. Standard methods of detecting lies are based on the physiological manifestations of the body. Classic detectors take into account the indicators of cardiac activity and respiration. In addition, experts pay attention to changes in the timbre of the voice and facial expressions. In contrast, the online polygraph calculates the tactics of conducting a dialogue and shows successful results in recognizing liars. The technology analyzes the messages written and on this it concludes whether the user is truthful or untruthful. To test the technology, the researchers organized a virtual game, the participants of which were divided into two groups. In one of them, everyone was “saints” and spoke only the truth. Another group consisted of "sinners" who deliberately utter lies. During the game, everyone corresponded, during which the device noticed what words were used in both groups and how liars and truth-tellers were building a conversation.
The study helped to reveal some very interesting patterns. It turned out that the members of the group "for the truth" and representatives of the "sinners" select completely different sets of words. At the same time, they also type their messages with different typing speeds. Participants of the "deceitful" answer faster. At the same time, they use vocabulary that will more emphasize their confidence, for example, the ultimatum words “always”, “never”, etc. Truth-seekers express more doubts in words, think about the answer longer, build consistent reasoning using the words “perhaps”, “ because "and" probably ".
As a result, the virtual lie detector, based on the speed of responses, the choice of words and their frequency, makes accurate conclusions from 85 to 100%. The developers admit that not many participants were involved in the study, but they are confident that their device has great potential. Their plans are to further improve the technology and its subsequent distribution using various instant messengers. The creators of the device hope that their online lie detector will become the basis of the first digital polygraph that will help protect users from online fraud. In the future, the developers want to bring the device to the market, but the exact dates have not yet been announced. The Topic of Article: The online polygraph has learned to identify liars by the way they speak. |
