The robot that knows when you’re lying: Scientists create an AI that can detect deception with 73% accuracy by measuring subtle facial movements 

- Advertisement -


  • Israeli researchers caught ‘liars’ with an accuracy of 73 percent
  • They achieved this score by measuring small movements of facial muscles.
  • New technology may one day serve as the basis for the development of cameras and software capable of detecting deception in many real-life scenarios

- Advertisement -

Being able to detect when someone is lying has been a goal for decades, and now, thanks to artificial intelligence, scientists believe they are getting closer.

In a very early-stage trial, a team from Tel Aviv University placed sensors on volunteers’ faces, and looked for changes in subtle facial movements as they told a lie or a truth.

advertisement

The system was able to tell whether a person was lying with 73 percent accuracy—slightly less than the polygraph test, which is 80 percent accurate.

However, the team says that this is a very early stage in its development, and it will improve in the future.

- Advertisement -

They predict that in the future AI-equipped cameras could be used at airports, in online job interviews, or in police suspect interviews to see if someone is fibbing.

Being able to detect when someone is lying has been a goal for decades, and now, thanks to artificial intelligence, a team from Israel believes they are getting closer.

How does this work?

The researchers attached stickers with their special electrodes to two groups of facial muscles:

  • cheek muscles close to the lips
  • muscles above the eyebrows

Participants were asked to face each other in pairs, wearing a pair of headphones through which the words ‘line’ or ‘tree’ were transmitted.

When the wearer heard ‘line’ but said ‘tree’ or vice versa he was clearly lying, and his partner’s job was to try to detect the lie.

Then the two subjects switched roles.

As expected, participants were unable to detect their partners’ lies with any statistical significance.

However, the electrical signals delivered by electrodes attached to their faces led to false identification at an unprecedented success rate of 73%.

advertisement

Previous research has found that humans can detect lies from truth about 55 percent of the time, and a polygraph machine is up to 80 percent accurate.

It’s not enough to accept the results of a polygraph test as evidence in court, so researchers around the world are working on new solutions.

The Tel Aviv project worked by using machine learning and artificial intelligence to quickly analyze very small changes in muscle movements while lying down.

This involved small movements of the cheek muscles and eyebrows.

Facial movements were measured using stickers printed on soft surfaces that contain electrodes capable of monitoring and measuring nerves and muscles.

Professor Levy said: ‘Many studies have shown that it is almost impossible for us to tell when someone is lying to us. Even experts, such as police interrogators, do this only slightly better than the rest of us.

‘Current lie detectors are so unreliable that their results are not admissible as evidence in courts of law – simply because one can learn to control their pulse and deceive the machine.

‘Consequently, there is a great need for more accurate deception-detection technology.

‘Our study is based on the assumption that facial muscles distort when we lie, and so far no electrode has been sensitive enough to measure these contortions.’

In the study, researchers attached stickers to the muscles of the cheeks close to the lips and the muscles above the browbones.

They had the volunteers sit in pairs facing each other, before the words ‘line’ or ‘tree’ were spoken through the headphones into their ears.

Volunteers were instructed to lie or tell the truth to their partner about what word they heard.

Researchers placed stickers with special electrodes on two groups of facial muscles, above the lip and above the eyebrow.  Volunteers again told lie or truth

The researchers placed stickers with special electrodes on two groups of facial muscles, above the lip and above the eyebrow. Volunteers again told lie or truth

We speak slower when we’re lying, study finds

According to a 2021 study, liars are more likely to speak slowly and put less emphasis in between words.

According to researchers in Paris, the brain can detect a signature in a liar’s voice — slower speech and less emphasis in the middle of a word.

This process occurs even when we are not actively trying to determine whether someone is honest or not.

It is hoped that the discovery could be used in the future to develop ‘lightweight tools’ that police can use to determine whether a criminal is lying.

Read more: Liars speak slowly and put less emphasis in between words

advertisement

When the wearer heard ‘line’ but said ‘tree’ or vice versa he was clearly lying, and his partner’s job was to try to detect the lie.

The results showed that participants were unable to detect their partners’ lies with any statistical significance.

However, the electrical signals picked up by the stickers on the face can be detected as lies 73 percent of the time.

‘Since this was a preliminary study, the lie itself was very simple. Usually when we lie in real life, we tell a long story that includes both a deceptive and a true component,’ Professor Levy said.

‘In our study we had the advantage of knowing what participants heard through the headset, and therefore also knowing when they were lying.

‘Thus, using advanced machine learning techniques, we trained our program to detect lies based on the EMG (electromyography) signals coming from the electrodes.

‘Applying this method, we achieved an accuracy of 73% – not perfect, but much better than any existing technology.

‘Another interesting finding was that people lie through different facial muscles: some lie with their cheek muscles up and others lie with their eyebrows.’

The researchers believe their AI will have ‘dramatic effects in many areas of our lives’.

They predict that in the future, AI-equipped cameras could be used at airports, in online job interviews, or in interviewing police suspects to see if anyone is fibbing.

He anticipates that AI-equipped cameras could be used at airports in the future.

,

- Advertisement -
Mail Us For  DMCA / Credit  Notice

Recent Articles

Stay on top - Get the daily news in your inbox

Related Stories