This company says it is developing a system that can only recognize your face from your DNA.

Parabon’s technology “doesn’t tell you the exact number of millimeters between the eyes or the relationship between the eyes, nose and mouth,” says Greytak. Without that kind of precision, facial recognition algorithms can’t produce accurate results, but getting such accurate DNA-based measurements would require fundamentally new scientific discoveries, she says, and “papers that tried to make predictions at that level didn’t have a lot of research.” success.” Greytak says that Parabon only predicts the general shape of someone’s face (although the scientific feasibility of such a prediction has also been questioned.).

Police are known to conduct forensic examinations based on witnesses’ descriptions through facial recognition systems. A 2019 Georgetown Law Center Privacy & Technology Survey found that at least half a dozen police agencies in the US “allow, if not encourage” the use of hand-drawn or computer-generated forensic sketches as input photographs for facial recognition systems. AI experts have warned that such a process is likely leads to a decrease in the level of accuracy.

Corsight has also been criticized in the past for exaggerating the capabilities and accuracy of its facial recognition system, which it calls “the most ethical facial recognition system for very challenging environments,” according to the slide. presentation available online. V technology demonstration for IPVM Last November, Corsight CEO Watts said that Corsight’s facial recognition system could “identify a person wearing a mask — not just a mask, but a ski mask.” IPVM reported that using Corsight’s AI on a masked face gave 65% confidence, Corsight’s own measure of how likely a captured face would be to match in its database, and noted that a mask is more accurately described as a balaclava or neck. gaiters, as opposed to a ski mask with cutouts for only the mouth and eyes.

Broader problems with the accuracy of facial recognition technology have been Welldocumented (including MIT Technology Review). They are more pronounced when photographs are poorly lit or taken at extreme angles, and also when subjects have darker skin, are women, or are very old or very young. Privacy advocates and the public are also critical of facial recognition technology, especially systems such as Clearview AI which scrape social media as part of their matching engine.

The use of this technology by law enforcement is especially dangerous – Boston, Minneapolis and San Francisco are among the many cities where it is banned. Amazon and Microsoft have stopped selling facial recognition products to police groups, and IBM has taken its facial recognition software off the market.


“The idea that you can create something with the level of detail and fidelity required to conduct a face-matching search is ridiculous to me,” says Albert Fox Kahn, civil rights lawyer and executive director of the Surveillance Technology Oversight Project. , which is actively working on issues related to facial recognition systems. – It’s pseudoscience.

Jemila Sero, Research Fellow Computational Visualization Group from Centrum Wiskunde & Informatica, the national research institute for mathematics and computer science in the Netherlands, says the science supporting such a system is still underdeveloped, at least publicly. Sero says the catalog of the genes needed to produce accurate images of faces from DNA samples is currently incomplete, citing a 2017 Human Longevity study.

In addition, factors such as environment and aging have a significant impact on faces that cannot be captured by DNA phenotyping, and studies have shown that individual genes do not affect the appearance of someone’s face as much as their gender. and origin. “Premature attempts to implement this technique will likely undermine the credibility and support of genomic research and will not benefit society,” she said in an email to MIT Technology Review.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button