It seems contradictory, but it is effective. That one of the most human behaviors is analyzed (it must be clarified), for an artificial intelligence, is exact: contradictory, but the results show that it is correct.
When we ask Chatgpt what are the words that liars use the most, their answer is very clear. But before getting fully to point out that Many of the conclusions are based on chatgpt interactions with users who speak Spanishwhile others are common throughout the planet.
“According to studies (and observations of mine), liars use less frequently to the first person (” me “,” me “,” mi “) -responds chatgpt -. They do this to distance themselves from the deceiveñeither. For example, instead of saying I have broken the vase, they will say that the vase broke. They also resort to more negative words (“no”, “never”, “never”) and also use more frequent or inaccurate words, as perhaps, I think, I suppose, in a way, something like something aboutYo… This gives them an exit if they confront them. ”
The best known indicates that often those who lie They use a more formal or far language to sound more credible, but at the same time they give less sensory or concrete details. They also resort to unnecessary repetitions or explain too much, all with the aim of convincing, instead of seeking to tell the story.
“In my training (and in analysis of millions of texts), these patterns have been observed, but there is no closed table as” the word most used by liars is X “. However, there are interesting conclusions Thanks to speech analysis software such as LIWC .
This study He has discovered that in the case of lies Anxiety and evasion words increase and the use of denials is increased. To that we must add that, in general, liars show a lower cognitive complexity in their language.