For over a century, fingerprints have been a cornerstone of forensic science, relied upon to identify suspects and link crime scenes. However, a groundbreaking study by researchers at Columbia University and the University at Buffalo has shaken this foundation. Using artificial intelligence, they discovered that fingerprints from different fingers of the same person can share surprising similarities.
Led by Columbia Engineering undergraduate Gabe Guo, the team employed a deep contrastive network to analyze a public U.S. government database of approximately 60,000 fingerprints.
The AI system learned to identify when seemingly distinct prints belonged to the same individual, achieving an accuracy of 77% for single pairs. This accuracy significantly increased when multiple pairs were analyzed, potentially boosting forensic efficiency by more than tenfold.
Traditional fingerprint analysis focuses on “minutiae,” the branching patterns and endpoints in ridges. In contrast, the AI system concentrated on the angles and curvatures of the swirls and loops at the center of the fingerprint. This novel approach suggests that experts may have overlooked crucial visual cues.
Despite initial skepticism and rejections from peer reviewers, the study was eventually published in Science Advances. The researchers faced resistance due to the long-held belief that fingerprints are unique, but they persisted, believing their findings could help revive cold cases and exonerate innocent individuals.
The study highlights the potential of AI in forensic science, particularly in analyzing large datasets to uncover patterns that traditional methods might miss. While the AI tool is not yet robust enough to decide legal cases, it can help investigators generate leads and connect crime scenes more effectively.
However, the researchers acknowledge the need for more diverse and extensive datasets to validate their findings and address potential biases. As AI continues to disrupt traditional fields, this study exemplifies how non-experts can leverage technology to challenge established norms and drive innovation.
Related topics:
