AI Discoveries Challenges The Belief In The Uniqueness of Intra-Person Fingerprints

By Columbia University School of Engineering and Applied Science

In a groundbreaking study led by Columbia Engineering undergraduate senior Gabe Guo, the long-standing belief in the uniqueness of fingerprints from different fingers of the same person, known as “intra-person fingerprints,” is brought into question. Traditionally considered the gold standard in forensics, fingerprints have been assumed to be unique and unmatchable within the same individual.

Guo’s team, in collaboration with Hod Lipson’s Creative Machines lab at Columbia Engineering and Wenyao Xu’s Embedded Sensors and Computing lab at the University at Buffalo, SUNY, utilized an artificial intelligence-based system called a deep contrastive network. By analyzing a public U.S. government database of 60,000 fingerprints, the AI system demonstrated an accuracy of 77% in determining whether seemingly unique fingerprints belonged to the same person. With multiple pairs, the accuracy significantly increased, potentially revolutionizing forensic efficiency.

The findings, challenging the conventional belief in fingerprint distinctiveness, were initially met with skepticism in the forensics community. Despite rejections from forensics journals, the team persevered, refining their AI system and submitting their work to a broader audience. The paper, now accepted by Science Advances, introduces a new forensic marker based on angles and curvatures rather than traditional minutiae.

While the system’s accuracy is not yet sufficient for official case decisions, it offers a valuable tool for prioritizing leads in ambiguous situations. Hod Lipson emphasizes the transformative potential of AI-led discoveries, asserting that even non-experts can challenge established beliefs. As AI continues to evolve, its impact on scientific discovery may be more far-reaching than previously imagined.