Scientists warn that new AI-generated faces are seen as more trustworthy than real ones

 

As if the notion that AI is already aware wasn't scary enough, researchers have shown that AI-generated faces have advanced to the point where many people believe they are more trustworthy than real people.

According to a news release from Lancaster University, a pair of researchers revealed that a neural network termed StyleGAN2 is capable of constructing faces that are indistinguishable from the real thing. In a surprising twist, participants tended to view AI-generated faces to be more trustworthy than real-life appearances.

"Our evaluation of the photo realism of AI-synthesized faces reveals that synthesis engines have crossed over the uncanny valley and are capable of making looks that are indistinguishable — and more trustworthy — than actual faces," the researchers said in a press statement.

 

Dark Implications

The researchers asked participants to score the trustworthiness of the faces on a scale of one to seven, with one being the least trustworthy. On average, the synthetic faces were thought to be 7% more trustworthy than real ones.

Of course, the team's discoveries have a slew of ominous consequences, including the possibility that it would pave the way for ever more advanced deepfaked images and films that are indistinguishable from reality. Worse, it raises the possibility that it may throw doubt on authentic images and recordings.

"Perhaps the most damaging effect is that in a digital world where any picture or video can be falsified, the validity of any uncomfortable or undesired recording might be questioned," the researchers said.