Corsight AI, an Israeli-founded organisation providing facial recognition services, claims to have devised a way to create facial profiles from DNA samples. The product was first introduced by Corsight CEO Robert Watts and executive vice president Ofer Ronen at the Imperial Capital Investors Conference in New York City on December 15.
The product roadmap included “voice to face,” “DNA to face,” and “movement” (or gait recognition) as expansions of the company’s facial recognition capabilities. According to a company slide deck made available to the surveillance research group IPVM, the “DNA to Face” product “constructs a physical profile by analysing genetic material contained in a DNA sample.”
The company has previously drawn flak over exaggerating the competency and accuracy of its facial recognition system. Last November, Corsight CEO Watts claimed that Corsight’s technology could “identify someone with a face mask—not just with a face mask, but with a ski mask.” However, Corsight’s AI only clocked a 65% confidence score when completing this task.
AIM Daily XO
Join our editors every weekday evening as they steer you through the most significant news of the day, introduce you to fresh perspectives, and provide unexpected moments of joy
Your newsletter subscriptions are subject to AIM Privacy Policy and Terms and Conditions.
The company keeps its work low-key and has not lifted the lid on its plans and future products. However, marketing materials suggest CorsightAI is targeting its services towards government and law enforcement agencies.
Old wine in new bottle
The “genomics-based, healthy intelligence” company Human Longevity had claimed to use DNA to predict faces back in 2017. However, experts found such claims to be suspect, and a former employee had attested that Human Longevity couldn’t pick a person out of a crowd using a DNA sample. Further, the chief science officer of the genealogy platform MyHeritage, Yaniv Erlich, published a study highlighting the major flaws in Human Longevity’s research.
Download our Mobile App
Parabon NanoLabs uses its product line, Snapshot, to give law enforcement physical depictions of people from genetic samples. The phenotypic characteristics (such as eye and skin colour) of these computer-generated renderings come with a confidence score. So, for example, there could be an 80% probability of the person being pursued having green eyes. According to Parabon’s director of bioinformatics, Ellen McRae Greytak, the company has helped solve over 200 cases in the last seven years.
Unlike Corsight, Parabon doesn’t claim the physical profiles they create can be used as input for facial recognition systems. The technology isn’t precise enough for facial recognition algorithms to deliver accurate results.
While experts claim the science to support Corsight AI’s product doesn’t exist, it could compound the ethical, privacy, and bias challenges facial recognition technology is already causing.
Predicting human physical traits from genomic data sparks privacy concerns. Albert Fox Cahn, a civil rights lawyer and executive director of the Surveillance Technology Oversight Project, claims the idea of creating something that has the “granularity and fidelity” to be put through a facial recognition system “is pseudoscience.”
Evolution
Facial recognition technology has gone over numerous changes since its inception. Coined in 1960, identities were automatically differentiated based on the manual marking of various “landmarks” on the face, like the placing of the eyes and the mouth. Later, the work was extended and standardised to include 21 specific subjective markers like hair colour and lip thickness to automate the recognition. In the late 80s, scientists applied linear algebra to the problem of facial recognition and formed the Eigenface system. It was in the early 1990s when development for the technology for commercial uses was initiated. In 2006, the US government supported the Face Recognition Grand Challenge (FRGC) to promote and advance face recognition technology. Here, 3D face scans, high-resolution face images, and iris images were used in the tests to make the technology 100 times more accurate.
It was not until 2010, when the consumer experienced face recognition technology that was introduced by Facebook to identify people whose faces featured in the photos of their users. The major breakthrough that we see now happened when Apple launched the iPhone X that could be unlocked with FaceID. Post that, the technology is being used by airlines, airports, border controls, stadiums, transport hubs, mega-events, concerts, and conferences, among others.
