Due to its uniqueness and permanence, fingerprint analysis is regarded as a valuable tool in the field of forensics and security because no two fingerprint patterns are identical, not even identical twins. There are so many unique aspects of fingerprints that even your own fingerprints do not match with the fingerprints of others.
However, there is new research utilizing artificial intelligence (AI) that claims that forensic fingerprint identification will undergo a substantial change.
There has been a recent breakthrough in forensics with the invention of a new artificial intelligence system developed by Columbia University engineers, which dispels a long-held belief in forensics: that fingerprints from a person's different fingers are not all the same.
In the field of forensics, it is widely accepted that fingerprints on different fingers of a given individual are all different from one another, and therefore, unmatchable. Columbia Engineering undergraduate student Gabe Guo led a team that challenged this widely held presumption and questioned this widely held belief.
With no prior forensic experience, Guo found a public database containing about 60,000 fingerprints and fed the fingerprints in pairs into an artificial intelligence-powered system called a deep contrastive network based on artificial intelligence. It is possible that the pair of fingers belong to the same individual at times (but with different fingers) and at other times it is likely that the pairs belong to different individuals.
Researchers suggest that the artificial intelligence tool was analyzing fingerprints differently from traditional methods - the algorithm focused more on the orientation of the ridges within a finger as well as how the end and fork of the ridge, which is referred to as minutiae, rather than the end of each ridge. Prof Lipson added that "there is no doubt that it is not using traditional forensic markers that have been used for decades," in the field of forensics.
As Prof Lipson explained, he and Gabe Guo, an undergraduate student, were both surprised when they saw the outcome of this particular experiment. It seems like something has been calculated based on the curvature and angle of the swirls in the centre.
During the Covid-19 lockdowns when Gabe Guo was stuck at home while waiting to start his freshman year at Columbia University, a professor asked Gabe Guo the following question during a casual chat. That conversation did not occur to me, but it set the stage for my life to be focused on it for three years to come.
A professor from the University of Buffalo, as well as one of Guo's coauthors, Wenyao Xu, conducted a study on the subject for Columbia's Department of Computer Science during his undergraduate years and Guo led the team.
Identical fingerprints don't exist everywhere on the planet, Guo and his colleagues argue, even though fingerprints have taken up a lot of space. This is apparently contrary to the long-established truth about fingerprints.
The paper is an interesting one, said Simon Cole, a professor in the department of criminology, law and society at the University of California, Irvine who believes its practical benefit is overstated as he agreed that it is interesting. This study was not conducted by Cole, nor was he involved in any of its components.
Essentially, Cole said that the fingerprint system could be useful at crime scenes where fingerprints found were from a different finger than those in police records, but Cole said that it would be a rare occurrence. When fingerprints are taken, all ten fingers and palms are routinely recorded when the prints are taken, so Cole said this could only happen in a few cases. It appears to me that authorities are not clear as to why they think law enforcement does not have all records of an individual's fingerprints or even some of them, he said.
Champion and Cole both praised the decision that the team behind the study made in open-sourcing the AI code so it could be checked by others, a decision which was praised by the team behind the study as well. Despite that, Guo seems to believe that the study is of much greater importance than just fingerprints alone.
According to Dr Hod Lipson, a Professor of Robotics at Columbia University, this activity is clearly different from the traditional forensic markings used by forensic detectives over many years. The curvature and the angle of the swirls in the middle seem to indicate that it uses something like that in the centre." However, the development comes as no surprise to an academic in England who has been working on this topic for over twenty years.
It has been reported by the BBC that Professor Graham Williams from Hull University has stated that determining what makes a fingerprint unique has never been an exact science. The uniqueness of fingerprints remains uncertain, as it is acknowledged that current knowledge is limited to the absence of demonstrated cases where two individuals possess identical fingerprints. The implications of this uncertainty call for further research, study, and deliberation. According to the report, the identified findings are not anticipated to exert a substantial influence on the field of forensics at this juncture.
