A group at the US college prepared a simulated intelligence device to look at 60,000 fingerprints to check whether it could sort out which ones had a place with a similar person.
The analysts guarantee the innovation could recognize, with 75-90% exactness, whether prints from various fingers came from one individual.
Be that as it may, they are don't know the way in which it works.
"We don't be aware without a doubt the way that the computer based intelligence makes it happen," conceded Prof Hod Lipson, a roboticist at Columbia College who managed the review.
Criminology
The scientists think the simulated intelligence device was breaking down the fingerprints another way to conventional strategies - zeroing in on the direction of the edges in the focal point of a finger as opposed to the manner by which the singular edges end and fork which is known as particulars.
"Obviously it isn't utilizing conventional markers that legal sciences have been utilizing for quite a long time," said Prof Lipson. "It appears as though it is utilizing something like the ebb and flow and the point of the whirls in the middle."
Prof Lipson said both he and Gabe Guo, an undergrad understudy, were both astonished by the result.
"We were extremely doubtful... we needed to check and twofold check," he said.
That may not be brand new information to others in the field.
Graham Williams, teacher of measurable science at Body College, said the possibility of one of a kind fingerprints had never been firmly established.
"We don't really realize that fingerprints are special," he said. "All we can manage to say is that to the extent that we know, no two individuals presently can't seem to exhibit similar fingerprints."
Crime locations
The consequences of Columbia College's review could can possibly influence both biometrics - utilizing one specific finger to open a gadget or give distinguishing proof - and measurable science.
If, for instance, a unidentified thumb print is found at crime location A, and a unidentified pointer print at crime location B, the two couldn't at present be forensically associated with a similar individual - yet the simulated intelligence device might recognize this.
The Columbia College group, none of whom have criminological foundations, conceded that more exploration was required.
Simulated intelligence apparatuses are regularly prepared on tremendous measures of information and a lot more fingerprints would be expected to foster this innovation further.
Furthermore, every one of the fingerprints used to foster the model were finished prints and of good quality, though frequently in reality fractional or unfortunate prints are bound to be found.
"Our apparatus isn't sufficient for concluding proof in legal disputes however it is really great for producing leads in criminology examinations," guaranteed Mr Guo.
Yet, Dr Sarah Fieldhouse, academic administrator of criminological science at Staffordshire College, said she didn't figure the review would have "huge effect" on criminal casework at this stage.
She said there were inquiries around whether the markers the simulated intelligence device was zeroing in on continued as before relying upon how the skin contorted as it came into contact with the print surface, and furthermore whether they continued as before throughout a lifetime, as customary markers do.
Yet, this could be interesting to reply as the scientists are unsure about precisely the thing the man-made intelligence is doing, just like with numerous artificial intelligence driven apparatuses.
The Columbia College concentrate on has been peer-evaluated and will be distributed in the diary Science Advances on Friday.
Yet, a couple of twins in Cheshire may be in front of everybody. Their grandma Ditty told the BBC her two grandkids can open each other's iPhones utilizing their own fingers.
"They showed me on Christmas day," she said. "We were informed they were indistinguishable when they were conceived yet I can differentiate between them as they have more seasoned."
She asserted that her grandkids can likewise sidestep the handsets' facial acknowledgment include.
Fingerprints are framed before birth. Research distributed last year recommended the hereditary cycle behind them might be like the manner in which creatures like zebras and panthers get their markings: a hypothesis previously proposed by codebreaker Alan Turing during the 1950s.
