California Blocks Police Body Cam Use of Face Recognition
aclu.org
aclu.org
I know that people here have a knee-jerk privacy big-brother is watching reaction, and I have that too. But beyond that I want to think if it was something that could have been beneficial to society or not.
The hard part is figuring out which ones are the criminals and which ones are politicians doing their thing.
It's both funny and sad at the same time on multiple levels.
First, the threshold used was the default suggested for one-to-one matching (lower false reject rate, higher false accept rate). It doesn't make sense to use this threshold for search applications, and when the study was reproduced with a more appropriate (higher) threshold there were 0 mis-identifications.
Second, law enforcement use of face recognition doesn't even involve the algorithm making a lights-out identification decision. Instead, the most similar faces are presented to the user in ranked order by similarity (like a search engine). It's a tool for generating investigative leads, often preferable to publishing the face image of a wanted perpetrator on local news.
What makes you think that a LEO would change the defaults? They're not computer scientists, they're people who bought a product to perform a specific function.
It's a tool for generating investigative leads
So were photo books, lie detectors, surveillance cameras, and other tools at first. They became a close enough proxy for "guilt" in the eyes of the users for people to automatically be treated as criminals, even if they'd done nothing wrong.
The analogy would be blaming an anchor manufacturer when the "default anchor" for rowboats doesn't hold an oil tanker.
What do you mean "mis"? It correctly identified politicians as criminals. In that, this AI showed itself to be lightyears ahead of the electorate in terms of understanding how politics works.
Either the politicians are criminals, and this tool was only able to correctly categorize 26 of them.
Or the politicians are not criminals, and this tool erroneously categorized 26 of them as such.
And that's based only on what we know. The experiment may have given rise in state leaders to more fundamental objections.
Consider, what if the 26 misidentified politicians, were largely black or hispanic males?
We just need to know more about what's been happening in all the testing to say whether these systems are useful or not. That's all I'm saying.
At least one body camera on the market now is fully capable of running FR software and has full networking built in - essentially a smartphone.
It scares me that the DHA has a picture of my face without my knowledge
I have not yet had to deal with that, but CBP tries very hard to capture your facial biometrics when you cross the US border. FYI, US citizens are not required to use those kiosks with the facial biometric scanners. You can simply present yourself directly to a CBP officer with your passport and completed Form 6059B and they must admit you.
The most recent time I entered the US, CBP insisted repeatedly that they had to use facial biometrics in order to assess the validity of my claim to US citizenship (i.e., to authenticate me against my passport) and that without submitting they would not admit me into the country. I refused, as CBP has neither the constitutional nor the statutory authority to seize my biometrics. In the end, the CBP supervisor went into his office, presumably called CBP lawyers, and then one particularly dull CBP officer did a literally minute long theatrical hexadecimal-take looking back and forth at my face and at my passport photo.
Again, CBP tries really hard to collect all the data they can, and they do everything possible, including outright lying and threatening, to make people think they just have to go along with it. If you're a US citizen, you don't, and I encourage you to join me in resisting their tyranny-building.
I know they do some good things, but they're definitely not hitting home runs when it comes to making the world a better place.
Do you not see the irony in that statement in the context of discussing whether further infringements on our civil liberties are worth it because "bad guys"?
So we should ban cash because it’s used for money laundering?
Now we might be able to use facial recognition to quickly find the robbers, but apparently that's bad.
[0]: https://www.reuters.com/article/us-usa-surveil-lance-watchdo...
freedom from disease and crime is not outlined anywhere in the constitution afaik...
It's hard to believe people would willingly say such foolish things, yet here we are.
That's demonstrably false. Additionally, law enforcement in the US has far too much power and has shown a proclivity to abuse every kind of power they have the opportunity to.
> be forced to deal in cash
can be reframed as "require cash as a payment option"
> forcing cities to allow slums on their streets
"prevent the forceful removal/evictions of tent cities"
> preventing automated detection of wanted criminals
"prevent mass-collection of anyone's face, not just criminals"
You're really going to freak out when you realize the government already has everyone's faces. Do you think they just throw away the photo after they create your ID?
https://www.theatlantic.com/ideas/archive/2018/11/aclu-devos...
Even if the system doesn't work super well now, systems like these improve as they are fed more data and told when they guessed wrong. That's why it's called "machine learning".
But obviously the required resources between these two solutions vary wildly.
There's a tendency to trust what comes out of a machine as infallible. I certainly don't want anyone armed carrying a badge getting notified in real-time of false-positives for potentially violent, armed, most-wanted criminals.
It just becomes a green light for use of excessive force. "The officer feared for his safety, the automated identification system reported the driver with the busted tail light as being the violent drug lord known as El Chapo. Our condolences go out to the substitute teacher's friends and family."
You're making a typical mistake by thinking these systems will not improve over time as they get more training data. They will, and given enough time, data, and elbow grease they will eventually reach superhuman levels of accuracy. But for them to get there they need to be exposed to the world under realistic circumstances.
With a few thousand officers doing this over the course of a year (collecting potentially a million training samples or more), the accuracy figures will go up tremendously, and the probability estimates will become far more accurate.
That's assuming that the body cam manufacturer has a capable deep learning research team, though, which might not be the case.