Nick Ferraro, Nick Faira, Thirty Million discussed on Malicious Life

Malicious Life
|

Automatic TRANSCRIPT

Blurry photos you drunkenly took in a low lit bar. It's not an easy thing to do. D- algorithm of it Are are a little bit difficult and you need to high quality photos. Unfortunately the quality of photos is becoming a lot a lot more improved over time. And of course you can. There can be photos of you tagged with your identity without you even doing it right on social media. Your friends. Photo said the availability of footage. Were you can be identified as has been significantly increased. Have you ever uploaded a drunken late night photo to facebook. But facebook still suggested the correct people to tag even before you could tag him. The algorithms have gotten really good in recent years. In many cases they beat humans. Two years ago for instance the national institute of standards and technology the nist gave a set of twenty pairs of images designed to be very difficult to parse to two groups the first group where leading facial recognition algorithms the second were humans. But not just humans experienced forensic examiners and i. St researcher summed up their findings succinctly quote. Well it turns out the best algorithm is comparable to the best humans and quote. There are so many cases of machines now surpassing humans in facial recognition. in fact. it's not even an you. Phenomena a decade and a half ago researchers began building algorithms that could surpass ability to recognize one another under specific conditions cold heartless machines. Even beat us at recognizing emotions in two thousand fourteen accompany founded out of the university of california at san diego developed an algorithm which could distinguish between when someone was making a genuine facial expression and when they were just acting at a rate of eighty five percent accuracy. Humans in that same test could tell only fifty five percent of the time in other words machines would hate movies. they'd see right through merrill strip. This isn't to say that all facial recognition is better than humans. Or even good at up. It really depends on which algorithms you choose last year. And i steep tested one hundred eighty nine different algorithms sold on the market developed by nine different companies. They conducted a variety of tests and suffice to say the results varied so any state Didn't experiment where the detroit to match up Well known public figures i think. D specifically tested congress people in the united states against a database of of criminals and got a surprisingly high match. So i think they got like thirty or forty false positives so these algorithms that they're not super accurate and they used a really really good Data source right. Because they were matching up against criminal photos. Criminal photos they're der der taking us well-formed portrait photos of people. Andy were matching them. Up against really nice well aligned. Photographs of congress be put the tent of an ice teas. Findings were shocking. Prompting calls for investigation by washington lawmakers. We don't have time to review the whole study but for a sense of just how bad some of the results were considered this. Some algorithms falsely identified saidan racial groups a hundred times more often than others hundred. That's an incredible and dangerous result for algorithmic which are currently on the market and being used out in the world today with care of your specifically we don't have access to to technology into their specific database to use. There's a reason to be suspicious about whether something like clear. View could actually work for one thing. It was largely built by one engineer. I still haven't gone over that part. Furthermore database of three billion images is really difficult to wrangle. Having lots of training data is always good in ai but heading to cross referenced. Three billion data points for any given query is a different story. So let's say. I look like a hundred people out of debt. You know billions live into world power. I look like more than one hundred people. So thick does hundred people all of these hundred during the database. Then there's a good chance that if you showed even a human might picture in a picture of any of those hundred people they would say. Oh yeah this guy. And it's this guy and it's just so the larger the database using the larger probability of mistake there is no publicly verifiable data on the efficiency of clear views algorithm. So we can only infer from circumstantial evidence like from the testimony of florida. Detective named nick faira who spoke to the new york. Times for years ferrero had relied on florida's statewide facial. Recognition program called the face analysis comparison and examination system or faces for short faces work because it was designed for use across law enforcement in the state leveraging database of thirty million floridians mugshots and dmv photos. You'll recall that these kinds of photos of the best kind. You can give a program like this but when ferrara tried out clearview it was no contest. Faces didn't touch the web but the clearview pulled from everywhere faces required. Clear straight on pictures but clearview could handle angles and even hats glasses and partially covered faces. When ferrara ferried photos from old dead end cases. Clearview gave him back over thirty new suspects to look into one. Tonne tat told. Cbs that his program runs at an accuracy. Rate of ninety nine point six percent. He told the new york times. It was seventy five percent. Whatever the exact figure is. It's definitely high because one has a lot of customers and the pay an absolute premium for his service. Nick ferraro's police department pays ten thousand dollars for a year's subscription ccording to. Cbs one's biggest clients around twenty five thousand dollars per year. You better bet that for.

Coming up next