Artificial intelligence appears to have helped revolutionize forensic science. It turns out that our knowledge of the uniqueness of the fingerprint was wrong. Everyone knows that every person has unique fingerprints. A lesser known fact, but very important from a forensic perspective, is that every finger of every person also has a unique fingerprint. Or rather, everyone thought so until now. It turned out that wasn’t quite the case.

Using artificial intelligence, a team led by Columbia Engineering senior Gabe Guo has proven that this decades-old belief is actually false.

Guo, who had no background in forensics, used a publicly available U.S. government database of nearly 60,000 fingerprints. He “fed” all these traces into the artificial intelligence system in pairs. Some pairs belonged to the same person but different fingers, and some belonged to different people.

It turns out that different fingerprints of the same person are, of course, not the same, but they have many common features. This is sufficient to identify two fingerprints from different fingers belonging to the same person.

The accuracy of such a comparison is not very high; 77% for a pair of fingers. In this it thrives in the presence of several pairs. This accuracy is not enough for the technology to be evaluated in court, but the study was based on just 60,000 prints. Columbia Engineering senior engineer Aniv Ray, PhD, notes that this study is just the beginning.

Imagine how well it would work if trained with millions of fingerprints instead of thousands.

The scientific article was published in the journal Science, but in fact the path to publication was not easy for the authors. When the team received the results, they sent them to a well-known forensic journal, but they were rejected after a few months. Moreover, with the statement that “it is well known that every fingerprint is unique”, it is impossible to detect similarity even if the fingerprints belong to the same person. That is, the critic, a scientist by nature, acted on the principle “it cannot be, because it cannot be” and, as we now know, made a fatal mistake.

After being rejected, the team provided more data to the AI, which confirmed their findings, but was rejected again. And as a result, the paper was accepted by the journal Science Advances.

But there are also skeptics. Christophe Champod, professor of criminology at the School of Criminal Justice at the University of Lausanne in Switzerland, claims that some nominal similarities in different fingerprints of the same person have been known since the beginning of criminology. Champoda believes that the authors of the study did not discover anything new and simply created a “storm in the cup” as they used now widely known artificial intelligence.