Psychologists warn:

Artificial intelligence is not a lie detector

German experts are warning against the use of artificial intelligence (AI) to identify lies. In the journal "Trends in Cognitive Sciences", psychologists Kristina Suchotzki from the University of Marburg and Matthias Gamer from the University of Würzburg cite interviews with people seeking entry at EU borders or suspects in criminal cases as examples.

Such technology has already been tested at the EU's external borders - in Greece, Latvia and Hungary - Suchotzki told the German Press Agency, referring to the European border control project iBorderCtrl.

In general, Suchotzki and Gamer cite various problems: AIs are a black box whose judgments cannot be understood from the outside and therefore cannot be verified. There is also a risk of false results and a lack of a theoretical foundation, they say.

"The use of artificial intelligence in lie detection is based on the assumption that it is possible to identify a clear indication or a combination of indications that point to deception," Suchotzki is quoted as saying in a press release on the study. Decades of research have not been able to identify such evidence.

"Pinocchio's nose does not exist outside of books," emphasizes the duo. There are no valid behavioral clues that can be used to reliably distinguish whether someone is telling the truth or not. This is why traditional lie detectors, which attempt to deduce the truthfulness of statements from physical reactions such as blood pressure, pulse and breathing, have long been considered unscientific, the psychologists emphasize.