An international research team has made some interesting discoveries regarding the efficacy of artificial intelligence (AI) in determining a person’s age.

For years, AI has been able to automatically estimate a person’s age based on their facial features. This technology could be used to determine if a person can legally enter a bar, or who can view age-restricted online content. Some social media platforms such as Instagram and Tik Tok have discussed using AI for age-verification purposes.

One of the greatest debates in the world of AI is whether technology can perform tasks more effectively than people, with mere mortals being prone to human-error.

Researchers from Western University and Ben-Gurion University of the Negev in Israel set out to answer this question. Their team tested 21 different AI age-estimation technologies between 2020 and 2021. AI performance was compared to that of 30 undergraduate students from Ben-Gurion.

Researchers claim that humans often misestimate a person’s age when guessing based on appearance alone. Previous studies from the same international team suggest that humans tend to overestimate the age of people with smiling faces compared to folks with neutral expressions. Their research has also shown that the accuracy of estimations decreases for older faces.

In this study, researchers found that AI is inclined to make the same mistakes.

“Our results showed AI is even less accurate and more biased than human observers when judging a person’s age – even though the overall pattern of errors and biases is similar,” said senior author Dr. Melvyn Goodale, the founding director of Western University’s Brain and Mind Institute.

AI overestimated the age of smiling faces even more than humans did. Compared to human observers, AI showed a sharper decrease in accuracy when guessing ages of older versus younger adults, females versus males, and smiling versus neutral faces.

“AI tended to exaggerate the aging effect of smiling for the faces of young adults, incorrectly estimating their age by as much as two and a half years. Interestingly, whereas in human observers, the aging effect of smiling is missing for middle-aged adult female faces, it was present in the AI systems,” said Carmel Sofer, an associate professor Ben-Gurion’s Department of Cognitive and Brain Sciences.

According to the researchers, the current findings only allow for speculation as to why these biases occur. Certain factors such as the photo sets used to train AI could have had an impact on performance.

“The pattern of errors and biases we observed could provide some insights for the design of more effective AI technology for age estimation from faces,” added lead author Tzvi Ganel, also hailing from Ben-Gurion’s department of cognitive and brain sciences.

The findings from this study were published in Scientific Reports.