This is one of the most useful artificial intelligence (AI) application I’ve seen in a while: a virtual border agent that screens passengers at the border and uses facial recognition and other technologies to detect and predict whether passengers are telling the truth or not.
The virtual agent research programm, called “Automated Virtual Agent for Truth Assessments in Real-Time” (or AVATAR in short), was funded by the Homeland Security Department. It’s currently being pilot tested at the border between Arizona and Mexico.
It works like shown on the video below:
Eye-tracking is among the many technologies used in AVATAR do realize the screening: eye movements or changes in voice, posture and impercetible movements of the face muscles.
This project is the result of common work of researchers at the university of San Diego and at the university of Arizona.
Not only is the perspective of having virtual agents promising in terms of operational optimization at borders (do you remember how long it takes to go through border control?) but in my opinion it’s also a step forward in terms of ethical screening. Human controls are based indeed on cues that may be sometimes unethical (Cathy O’Neil has shown in her book that stop-and-frisk policies were unfair). Automated control with virtual agents promises a new era where all passengers will be equal in terms of treatment. The cues used for tracking will be objective and analysed only to predict risk. Borders will also be made much safer thanks to the possibility to compare passengers’ faces with those of criminals stored in central databases.
Image : shutterstock