By Angelina Tang
On February 2, 2024, a name likely familiar to many of you appeared on the radio show Science Friday. Gabe Guo, a current undergraduate senior at Columbia University majoring in Computer Science, published a research paper titled “Unveiling intra-person fingerprint similarity via deep contrastive learning” in the Science Advances Journal, January 12, 2024, alongside contributors Aniv Ray, Miles Izydorczak, Judah Goldfeder, Hod Lipson, and Wenyao Xu. In his research, it was found that, using AI analysis of fingerprints, there is a significant level of similarity between the prints on all of one individual’s fingers, contrary to prior belief.
For decades, it has been assumed in the forensic and medical communities that every single fingerprint on a person’s hand is unique from one another. However, Guo’s research sought to upend that assumption, inspired by a conversation he and Professor Wenyao Xu at the University at Buffalo had during the COVID pandemic timeframe, in which Professor Xu offhandedly brought up the concept of fingerprints from one individual being somehow similar–retrospectively, the idea makes logical sense. Back at Columbia University, with the help of Professor Hod Lipson, Guo decided to train AI to find similarities in fingerprints and identify whether or not two different prints belonged to the same person. He used a large bank of fingerprints from a U.S. government database, feeding it to the AI, and over time, the AI became very accurate at discerning subtle similarities—namely, the angle of the ridges near the center of the swirling fingerprint, the singularity—between prints belonging to the same person.
Guo fought against heavy resistance in the scientific community, as he failed to publish his research numerous times due to other scientists simply turning away and refusing to accept evidence contrary to the status quo. However, as he continued feeding the AI fingerprints and the data became more and more indisputably clear for his hypothesis, people began to take note. Following the project’s publication in Science Advances, he has been contacted by law enforcement agencies requesting to experiment with and apply this technique to their investigations, and he is also working to patent it. He said on Science Friday that as of right now, it’s not precise or professional enough to use as evidence in court, but it’s helpful for narrowing down suspects, as mentioned in the research paper. He says, “I’m sure that if the FBI decides to train this with their millions of fingerprints, they probably could get it to a level where it might be used as evidence. But I’ll wait for them to make that call.”
As Guo intends to pursue a PhD in Computer Science following his graduation in May, he intends to continue using AI to push its limits, destroying the preconception that AI can only do simple tasks and recall information. As he puts it, “We’ve built an AI that not only discovers new things, but knows our own bodies better than we do.” He says that he is interested in utilizing AI to study fields like crystallography, dynamical systems analysis, and precision medicine. In addition, when Gabe Guo was a student at East, he was the ESN Editor in Chief for two consecutive years. “I have to thank the teachers at East High School for putting me on the path to discover this,” he says. “In particular, Mr. Belling taught me to think like an engineer, and that science can be entertaining and even funny. Moreover, Mr. Huber and Mrs. Lanzone at the ESN played a huge role in nurturing my creativity and writing ability: although it’s not always taught in high school, the higher levels of science are really about creativity and communication.”
Guo’s research is a breakthrough for forensic science and can greatly improve investigation efficiency in finding suspects. In addition, it has also destroyed our preconceived notions about AI’s limitations and will be the gateway to further research in the medical field using it. Despite current fears of AI, we should look forward to more breakthroughs in science in the coming years thanks to proper use of AI as a tool to enhance research and unveil discoveries we would never have made on our own.
