With artificial intelligence, it’s not easy to distinguish the real from the fake. Especially when this technology is used to generate ultra-realistic faces. Researchers from the University of Lancaster have found that it is extremely difficult for us to differentiate a real face from another created from scratch by algorithms.
They conducted several experiments to determine how realistic these synthetic faces are. The researchers first asked 315 participants to identify, among 800 portraits, the faces of real people and those generated by artificial intelligence. They obtained an accuracy rate of 48.2%, slightly lower than if picking by chance.
The second group of participants were trained to recognize artificial faces before taking the same test. Their result? 59% accuracy.
ALSO READ: Valentine or swindler? Online dating and fraud on the rise
These mediocre scores are not surprising given the sophistication of the artificial intelligence software used to create ultra-realistic faces in just a few clicks. They are so good that some people use them for entertainment or parody purposes.
One of them had fun making Elon Musk appear in Stanley Kubrick’s film, 2001: A Space Odyssey, while another created a deep fake of Barack Obama calling Donald Trump a “dips***.”
This type of scenario is getting a lot of attention and raises many questions about the use of deep fakes to spread false information. Researchers at Lancaster University wanted to test whether synthetic faces elicit a sense of trust in viewers. They asked 223 participants to rate the trustworthiness of real and synthetic faces by giving them a score from 1 to 7.
The faces generated by artificial intelligence are perceived, on average, as 7.7% more trustworthy than human faces. A disturbing phenomenon that the researchers explained by the fact that the synthetic faces were quite “average” looking. This impression of being ordinary inspires a feeling of trust in most people.
“Our evaluation of the photorealism of AI-synthesized faces indicates that synthesis engines have passed through the uncanny valley [a theory by roboticist Masahiro Mori that affinity for robots increases with their resemblance to humans, ed. note] and are capable of creating faces that are indistinguishable — and more trustworthy — than real faces,” reads the study, which was recently published in the scientific journal PNAS.
So caution is the watchword when you come across a photo or a video on the internet. Especially those created with the Chinese application Zao. It allows its users to insert, in a few seconds, their face in film or video clips with a simple selfie. The result is sometimes impressively realistic. Proof that, on the internet, it is more important than ever to be wary of appearances.