How To Use Artificial Intelligence To Tell If A Person Is Lying

How To Use Artificial Intelligence To Tell If A Person Is Lying

  • 4
  •  
  •  
  •  
  •  
  • 3
  •  
  •  
  •  
  •  
  •  
  •  
  •  
  •  
    7
    Shares

Artificial intelligence, or AI, can tell if an individual is lying. Research began because although an individual’s identity can be confirmed fairly accurately with a photo ID, the same principle is invalid when the person is completing an online transaction. The question was if AI could establish when a person was lying by the way they moved their mouse.

The research involved a group of volunteers who were asked to either be truthful about who they were, or to memorize a false identity. The group was asked easy questions, with more complex questions mixed in. The idea was to catch the liars off guard because although information like a name or birthdate could be memorized easily, this information by itself was incomplete. If the individual was really born in California, but was pretending to be from Kansas, they would have to think when asked their home states capital. AI was able to identify these pauses by the movement of the cursor.

Additional experiments revealed when the algorithm data from the mouse path of the subjects was analyzed by AI, 95 percent of the liars were identified. When an individual is telling the truth, the line is a little cleaner, and AI is able to correctly interpret this data. If the accuracy can be improved, it is possible AI will be the latest weapon against identity theft.

Another important attribute of AI is it is capable of discerning what is missed by the human eye. AI can find the correlations previously undetectable by analyzing the corresponding data. This has never been possible when the process was accomplished manually. Research into AI in regards to determining the truth of an individual statement in a court of law or a trial is currently being conducted. There is a new type of artificial intelligence engine being used for covert automated deception detection. The engine is called the Deception Analysis Reasoning Engine, or DARE for short. This technology has the capability of detecting nuances in the human micro-expression by using video features.

A study from Dartmouth College and the University of Maryland showed deception could be determined in the micro-expression of the individual. The researchers used a system containing videos from actual courtroom trials. The process involved three basic steps. First, the audio was captured, then the video, then the data encoding. An AI engine was trained to recognize and identify the five most common micro-expressions. These are raising the eyebrows, frowning, turning the head to the side, protruding the lips, and turning the corners of the lips upwards. A database was then created by the researchers with video clips from certain trials. Fifty of these videos were truthful, and 54 were deceptive.

Fifteen people were asked to watch these videos to determine when someone in the video was telling a lie. When the same videos were interpreted using the DARE system, the rate of deception successfully detected was 92 percent. Artificial intelligence has the capability of combining video, audio and transcriptions. This means a determination can be made in real-time. AI can see all the different elements simultaneously, enabling the technology to determine which statements are true or false with an extremely high accuracy level.

AI is also playing an important role in security by assisting law enforcement agents in spotting deception. An artificial intelligence based kiosk had been specifically designed to detect travelers who are telling lies. This new lie detector was designed at San Diego State University by assistant professor Aaron Elkins. The kiosk has been named AVATAR, and was created to be used predominately at airports. When the individual has finished scanning their passport or ID, the kiosk asks the traveler a series of questions. The questions mix basic and complex and basic questions to determine if the person is lying.

This is accomplished by the system using sensors to analyze the data given by the body. This includes voice, tone, eye movement, posture, facial expressions and the dilatation of the pupils. The belief is it requires more effort to lie than to tell the truth. The additional effort is revealed through physical movements, and detected by AI. Although the technology of AI is still being consistently improved, there is no doubt the data it currently detects is accurate.

About the Author

Noelle Neff is a professional photographer covering niche markets. She graduated from the University of Miami and is based in the same city. Noelle Neff is also a renowned blogger on topics such as photography and travel. You can also find some of her shots at various online stock photo sites.