For over one hundred years, we've rested on the fact that every fingerprint is totally and completely unique, even on the same hand of the same person.
However, that commonly held belief has been called into question by recent research out of Columbia University.
Gabe Guo led a team trying to see if AI could detect new information from fingertips.
And it turns out that it can, at least if those prints are from the same person:
We show above 99.99% confidence that fingerprints from different fingers of the same person share very strong similarities.
And that's big news because sometimes the only fingerprints on record are from certain fingers and not others.
The AI looks at the swirl patterns and curvatures where the print lines begin, rather than the unique endpoints or the lines themselves.
Having another way to verify fingerprints is an especially big deal given a recent New Jersey ruling from the appellate court, which raised the bar on what prosecutors could use as fingerprint evidence.
The court brought up
A study commissioned by the Obama Administration that showed the error rate was ‘one in 306' and another study found the error rate was ‘one in [eighteen],' suggesting that ‘approximately [5%] or more of fingerprint identifications are false positives.'
The problem in this specific case was that the defendant's fingerprints were really, really close matches to the ones at the scene of a burglary, so close that for the past one hundred years, he would have gone to jail, but they weren't exactly, 100% the same as the ones on file, so the appellate kicked the case back to the lower court to reconsider.
And you can just hear every criminal in New Jersey calling their lawyer like
If this new technique really does add a new layer of fingerprint identification, it could clear up a bunch of the doubts in the 5% of false positives.
I've got to say, I'm a little concerned about the future this sort of research is creating, where we one day hear:
The computer finds you guilty on all counts!
P.S. Now check out our latest video 👇
