The Blurring Reality: Distinguishing Between Real and AI-Generated Faces in 2024

In an era defined by rapid technological advancements, artificial intelligence (AI) is redefining our understanding of what’s possible. From healthcare and finance to entertainment and art, AI’s capabilities are leaving an indelible mark on various domains. One area where AI has made significant strides is the generation of realistic images, including human faces. With the advent of powerful AI tools like Dall-E and Midjourney, creating lifelike images of people who don’t exist has become a reality.

The Challenge of Distinguishing Real from AI-Generated Faces:

The task of distinguishing between real and AI-generated faces has become increasingly challenging as AI systems have become adept at producing hyper-realistic images. Research conducted in recent years has revealed a startling phenomenon: AI-generated faces of white people are often perceived as more realistic than genuine photographs of white people. This phenomenon, known as hyper-realism, has raised concerns among experts and researchers alike.

Factors Influencing Perception of Realism:

Studies have identified several factors that influence our perception of realism when it comes to AI-generated faces. One key factor is the training data used to develop the AI system. AI tools trained on vast datasets containing primarily white faces tend to produce hyper-realistic white faces, while the realism of nonwhite faces may be less pronounced. This disparity highlights the need for more diverse training datasets to ensure that AI systems can generate realistic images of people from all backgrounds.

Another factor that contributes to the challenge of distinguishing real from AI-generated faces is the tendency of AI systems to create faces that conform to average proportions. These faces may lack distinctive features or imperfections that are commonly found in real people. As a result, participants in studies often fixate on deviations from average proportions, such as a misshapen ear or a larger-than-average nose, as signs of AI involvement.

Implications for Online Misinformation:

The ability of AI systems to generate hyper-realistic faces has significant implications for the spread of misinformation online. Digital fakes can be easily disseminated through social media platforms and other online channels, potentially misleading unsuspecting individuals. The confusion surrounding AI-generated faces makes it difficult for users to discern the authenticity of online content, potentially leading to the spread of false and misleading narratives.

Conclusion:

The ability of AI systems to generate hyper-realistic faces presents both opportunities and challenges. While these tools have the potential to revolutionize industries such as entertainment and fashion, they also raise concerns about the potential for misuse and the spread of misinformation. As AI technology continues to advance, it is crucial to develop strategies to mitigate these risks and ensure that AI-generated content is used responsibly and ethically.

Test Yourself: Which Faces Were Made by A.I.?

See if you can identify which of these images are real people and which are AI-generated.

Was this made by A.I.? 1 / 10
A.I. Real
How did you do? Were you surprised by your results? You guessed 0 times and got 0 correct.

Ever since the public release of tools like Dall-E and Midjourney in the past couple of years, the AI-generated images they’ve produced have stoked confusion about breaking news, fashion trends, and Taylor Swift.

Distinguishing between a real versus an AI-generated face has proved especially confounding.

Research published across multiple studies found that faces of white people created by AI systems were perceived as more realistic than genuine photographs of white people, a phenomenon called hyper-realism.

Researchers believe AI tools excel at producing hyper-realistic faces because they were trained on tens of thousands of images of real people. Those training datasets contained images of mostly white people, resulting in hyper-realistic white faces. (The over-reliance on images of white people to train AI is a known problem in the tech industry.)

The confusion among participants was less apparent among nonwhite faces, researchers found.

Participants were also asked to indicate how sure they were in their selections, and researchers found that higher confidence correlated with a higher chance of being wrong.

“We were very surprised to see the level of over-confidence that was coming through,” said Dr. Amy Dawel, an associate professor at Australian National University, who was an author on two of the studies.

“It points to the thinking styles that make us more vulnerable on the internet and more vulnerable to misinformation,” she added.

Top photos identified as “real” in the study
A.I.

93% got it wrong
A.I.

92% got it wrong
A.I.

90% got it wrong
Real

90% got it right
A.I.

89% got it wrong
Top photos identified as “A.I.” in the study
Real

90% got it wrong
Real

86% got it wrong
Real

84% got it wrong
A.I.

82% got it right
Real

79% got it wrong

The idea that AI-generated faces could be deemed more authentic than actual people startled experts like Dr. Dawel, who fear that digital fakes could help the spread of false and misleading messages online.

AI systems had been capable of producing photorealistic faces for years, though there were typically telltale signs that the images were not real. AI systems struggled to create ears that looked like mirror images of each other, for example, or eyes that looked in the same direction.

But as the systems have advanced, the tools have become better at creating faces.

The hyper-realistic faces used in the studies tended to be less distinctive, researchers said, and hewed so closely to average proportions that they failed to trigger suspicion among the participants. And when participants looked at real pictures of people, they seemed to fixate on features that drifted from average proportions — such as a misshapen ear or larger-than-average nose — considering them a sign of AI involvement.

The images in the study came from StyleGAN2, an image model trained on a public repository of photographs containing 69 percent white faces.

Study participants said they relied on a few features to make their decisions, including how proportional the faces were, the appearance of skin, wrinkles, and facial features like eyes.