AI Analysis Finds Fingerprints Less Unique Than Long Believed
-
A Columbia University senior used AI to analyze fingerprints and found similarities between prints from different fingers on the same person, challenging the belief fingerprints are unique.
-
The study relied on a deep contrastive network AI model to analyze a database of 60,000 fingerprint pairs, finding it could predict with 77% if prints were from the same person.
-
The AI focused on angles and curves in fingerprint centers to identify patterns. The lead researcher says findings have broad AI implications beyond forensics.
-
Study authors say it could help generate new leads in cold cases by matching prints. But accuracy isn't yet high enough to affect court cases.
-
The forensics community pushed back on the study, saying it doesn't radically challenge existing beliefs about prints. Critics say similarities between prints on same person are already known.