AI Surpasses Reality: AI-Generated White Faces Deemed More Convincing Than Real Humans.

Recent research led by experts at The Australian National University (ANU) reveals that artificial intelligence (AI)-generated white faces now appear more realistic than actual human faces, raising concerns about potential racial bias and misinformation in AI technology.

According to the study, more individuals perceived AI-generated white faces as human compared to real human faces, a trend not observed in images of people of color. Dr. Amy Dawel, the senior author of the paper, attributes this discrepancy to the disproportionate training of AI algorithms on white faces.

The potential consequences of consistently perceiving white AI faces as more realistic are significant, particularly in reinforcing racial biases online, Dr. Dawel warns. This issue is already evident in current AI technologies used for creating professional headshots, where AI may alter the skin and eye color of people of color to resemble those of white individuals.

One concerning aspect of AI "hyper-realism" highlighted by the researchers is that individuals often fail to recognize when they are being deceived. The study found that those who thought AI faces were real were paradoxically more confident in their judgments, indicating that individuals mistaking AI-generated faces for real ones may be unaware of the deception.

The researchers also identified specific reasons why AI faces can deceive people. Despite existing physical differences between AI and human faces, individuals tend to misinterpret them. For instance, white AI faces often exhibit more proportional features, leading people to mistake this as a sign of humanness. However, the researchers caution that relying on such physical cues may become unreliable as AI technology continues to advance rapidly, potentially eliminating visible differences between AI and human faces.

The researchers emphasize the need for greater transparency around AI technology to identify issues before they escalate into major problems. They highlight the potential risks, including the spread of misinformation and identity theft, and advocate for public awareness to mitigate these dangers. Educating the public about the perceived realism of AI faces, they argue, can foster appropriate skepticism about images encountered online.

The study, titled "AI Hyperrealism: Why AI Faces Are Perceived as More Real Than Human Ones," was published in Psychological Science, a journal of the Association for Psychological Science, on November 12, 2023.

Source: https://scitechdaily.com/ai-surpasses-reality-white-ai-faces-more-convincing-than-real-humans/

Comments

Popular posts from this blog

The highest court in Colorado has decided that Trump is not qualified to participate in the presidential race and has consequently excluded him from the state's ballot.

The stock market's upward momentum came to a halt as the S&P 500 experienced its most challenging day since October.