Due to rapid progress in artificial intelligencecomputer generated images, including hyper realistic faces – are becoming more common. Many of them look so convincing that they are difficult to distinguish from real photographs.
In a new studyResearchers tested people's ability to distinguish images of real faces from images generated by artificial intelligence and found that most participants missed most of the AI-generated faces. Even “super recognizers”—an elite group with exceptionally strong facial processing abilities—were able to correctly identify fake faces as fake only 41% of the time. Typical recognizers correctly identified only about 30% of AI-generated faces.
However, the study also found that people were better at detecting fake faces if they were given just five minutes of training beforehand. During the training, participants learned to identify common computer rendering errors, such as unnatural-looking skin textures or oddities in the way hair lies on the face. After training, detection accuracy increased significantly: super recognizers recognized 64% of fake faces, and typical recognizers – 51%.
Considering how difficult the task was even for highly skilled participants, how confident are you in your ability to recognize AI faces? Answer our poll below and let us know why in the comments.
—“I trust AI like a sailor trusts the sea. It can take you far or drown you”: survey results show that the majority do not trust AI
—“It’s impossible to put the genie back in the bottle”: readers believe it’s too late to stop the development of AI
—'Honestly, I'm not sure about it at all': Poll reveals public hesitation about experiments on conscious, lab-grown 'mini-brains'
