24.2 C
New York

AI researchers find our fingerprints may not be unique


A group of scientists has employed artificial intelligence (AI) to study a large number of fingerprints and has discovered that our fingerprints may not be as distinct as originally assumed, according to a report by BBC News.

From biometrics for unlocking mobile devices to law enforcement, fingerprints have been widely relied upon as a reliable means of identifying individuals, but this latest discovery may challenge that notion.

This US-based research conducted at Columbia University sought to determine if an AI system could accurately identify which prints belonged to the same person from a collection of 60,000 prints.

The findings indicate that the technology was able to match prints with an accuracy rate of 75-90%, although the methods it employed to achieve this are not yet completely understood.

It is possible that the AI system is taking a different approach to analyzing prints compared to traditional methods, focusing more on the ridges in the center of a finger rather than on minutiae such as the way individual ridges end or split.

Contested Perspectives

Professor Hod Lipson, a Roboticist at Columbia, commented on the unexpected results: “It’s evident that it’s not relying on conventional markers that have been used by forensic experts for decades. It appears to be using factors like the curvature and the angle of the swirls in the center.”

However, this revelation did not surprise Graham Williams, a professor of forensic science at Hull University in England.

Williams told the BBC that the uniqueness of fingerprints had never been definitively established. “We can’t assert that fingerprints are inherently unique. All we can say is that as far as we know, no two people have yet demonstrated the same fingerprints.”

These findings will likely spark further research, study, and debate. The report suggests that, at this stage, these revelations will not have a significant impact on the field of forensics.

The Columbia University study, which has undergone peer review, is set for release in the journal Science Advances on Friday.

Image credit, cottonbro studio, pexels.com

Related articles

Recent articles