Perhaps in the future the Internet will be cleared from the scammers, and by users will have to carefully watch their words. Researchers from Florida state University have developed artificial intelligence that perfectly fulfills the role of the polygraph detects the lie, watching over the ship the user with words and their speed of set. The test results showed that he was already able to distinguish truth from falsehood in 85% of cases, while the accuracy of these lie detectors varies only in the area of 60-70%.
To train the system to separate the true from the false, the developers have resorted to machine learning. The volunteer group of forty people were invited to play in the online game, in which they were randomly given the role of “Saint” or “sinner”. Over 80 gaming sessions they answered questions, and artificial intelligence to carefully monitor their messages.
It turned out that the “saints” or “sinners” and use completely different sets of words and print them at different speeds. For example, liars often use the words “always” and “never” to emphasize his confidence and respond very quickly. The people who speak the truth, often use the word “may” and words gain a more thoughtful and, consequently, slow.
In the end, paying attention to the frequency of words and speed of response, online polygraph learned to recognize the truth with accuracy from 85% to 100%. The researchers acknowledge that the test was attended by too few people, but even now it is clear that technology has a great potential. They intend to improve it and to contact the developers of various messengers to promote his brainchild.
This baseline study showed a great potential of developing a system of online printing to improve the communication in the Network. I want to draw the world’s attention to the idea that we could turn it into a commercial product that could be introduced to the messengers.
The shuyuan edition Ho, head of research
If you want such a system in instant messengers and other methods of communication on the Internet? How often do you face lies in the Network? Your answers can be left in the comments or in our Telegram chat.