it is only a matter of time until people start asking for more facial recognition. as a whole facial recognition has a bad reputation. but if you ask people if they like walking out of an airport without waiting in line at customs they will say yes. when it helps identify criminals who would otherwise have gotten away they will support it. over time it will become normalized
When it works, I'm sure it's great. Now consider being a non-white person and it misidentifies you where you wind up being arrested for something you didn't do.
Until it works equally well for everyone, it should not be used for things other than processes of convenience. Even then, it should have a way to gracefully fall back to something else that does not prevent anything
I would assume that as soon as it's introduced, the agents in charge of fallback will be lazy, as has already been proven from numerous reports of people (mostly black) being arrested from a false positive in facial recognition[1]
It depends on what it is. We've already seen situations where police arrested someone based on a facial recognition system incorrectly identifying someone. So, where's the supposed fallback in your opinion?
We've seen stories out of NYC where landlords have installed facial rec to access to tenants that didn't go so well. I can't remember if traditional key access was still available or not, but you could see where that would go bad if not.
The issue isn’t “is there a fallback” it’s “there is incredible bias in who has to use those fallbacks, who has to schedule trips allowing for that fallback, and who gets erroneously subjected to false imprisonment”.
If all people regardless of race, gender, or appearance were equally subject to the need to use the fallback then there would be less of an issue. But so far every commercial facial recognition program has demonstrated bias such that false positives and false negatives are much more common for PoC.
I opted out of this. There is insufficient transparency around the systems that process that data, who has access to it, what kinds of problems have occurred, and if it is being employed ethically (see the sibling comment about chronically poor facial detection of non-white people).
The degree of security theatre at airports aside, surely you don’t think public transit should suffer TSA-style security to warrant a solution to that self-imposed problem?
I'm always going to be against piling on significantly more invasive tools to make the state's job easier. I wonder if there's a word for this process where organizations make something intentionally difficult for us and then say "look at how hard this is, clearly you need to let me do this other thing to make it easier".
Also, one would have to assume that this function would be contracted out to a private party, which would sell the geolocation data of millions, or at least "share it with partners," because there is no US law against that.
None of the articles I've seen ever seen to address why NYC has such a high rate of fare evasion. Is it just an ingrained "cultural" thing now?
You could definitely jump over the TfL barriers if you wanted to, but no one does
> facial recognition seems like a fairly efficient way
I feel this is extraordinary dependent on how you measure "efficient" and how many false-positives or false-negatives are too many. (Also the fun problems where it's wrong way more often for certain groups of people.)
If the metric is solely "maximize the percentage of evaders who are punished", then the most "efficient" solution is to indiscriminately fine every single person. :p
It is if your idea of an "efficient" system is one with baked-in discrimination and lots of false positives. Targeting, not so coincidentally, folks in lower income brackets.
Because the city officials don’t want to draw attention to the racial demographics of fare evaders.
If they actually arrest them, then there would be official numbers that align with the anecdata that anyone who has ridden that system is already familiar with.
No, the issue is not “mugshots” bad it’s “automatic enforcement mechanisms with a very well documented tendency to produce false positives” are bad.
The issue is not “facial recognition to prevent fare dodging”, it’s facial recognition to provide “reasonable suspicion” to warrant arrest and detainment of specific groups. NYC has a long documented history of creating laws solely for the purpose of racial discrimination - “stop and frisk” laws that functionally only applied to black residents for example, IIRC it was only the lack of probable cause that did anything to limit that. Systems like this launder the racism and provide “probable cause” for unconstitutional search and seizure, as well as false imprisonment.
It also means you get subject to the stigma of being stopped and questioned by police when entering a station while the wrong race - it’s obviously worse for black Americans whom the police routinely execute because they’re scared (specifically racist claims of super human strength, claims the victims are substantially larger than they are, etc), but these systems mean you’re more likely to be falsely accused if you’re any group other than white with european descent. So obviously black people are going to be victimized here, but Pacific Islanders, Hispanic, more or less any Asian country, etc are demonstrably subject to higher false positive rates, and there’s no penalty for the false accusations or detention.
Yet they can setup automatic license plate readers to enforce "congestion pricing." Once more, those who follow the rules are punished, in favor of those who don't.
Apples to oranges comparison. There's a reason why we never accepted that humans should wear an identifying number while walking around the streets but we're happy for cars to do this.
Facial recognition software is equivalent to humans wearing a license plate. It's never been acceptable before and it's not acceptable now.
The MTA can't catch fare evaders because facial recognition is bad.
The NYC government can read peoples license plates and charge them arbitrary prices, partially, because the MTA is so broke.
So, fare evaders continue to get a free rides, in exchange for similarly Orwellian recording and logging of all vehicles entering and exiting certain parts of the city, which is effectively only one layer of abstraction different than scanning faces. Apples to pears, really.
It's what they do with the data in the long term that should be concerning in both cases.
Then I guess getting a plumber or an electrician to fix problems is a luxury item for everyone in NYC now too. Toll roads are not the exclusive route to and from a given area yet this system is. The tolls aren't even funding the road they're funding a broken public transportation system. There's a lot that's new here which is why this exact thing is often in the news.
Its not a "yet", its an effort that is part of a larger one. LIDAR and facial recognition should not be used. We should be against both.
Congestion pricing hits the lowest earning individual the highest. Congestion pricing should follow measurable improvements to public transit, focusing on the word "improvements".
Just set the fees to be a % of income + % of net worth. If you want to skip the complexity, making it a % of the current KBB value of the car that earned the ticket will be >99% just.
A few rich people driving shitboxes around Midtown to avoid expensive fees sounds like a fine outcome to me.
> Congestion pricing hits the lowest earning individual the highest.
This is true for anything that you buy. I’m not sure why being able to drive a car into the Manhattan CBD deserves special treatment. It also already has an excellent subway system that millions take daily.
> Congestion pricing hits the lowest earning individual the highest.
Can you tell me more? In most places, that does seem obvious. In NYC, it seems less obvious to me.
The friends of mine who complaining about congestion pricing rules are the upper/upper middle class folks who have not been on public transportation since Bloomberg’s first term. My friend who are lower income use public transportation everywhere.
In NYC there's a lot of public transportation however we're seeing its proposal in LA and other poor public transportation cities. In spaces where you need a car to get around it can be very hard. A bus here in southern california can take an hour when the drive takes 15 minutes or less.
The problem is that that’s essentially what most “facial recognition” systems used for law enforcement and adjacent do: they filter by skin color - false positives are overwhelmingly black.
“There has long been a concern [facial recognition] could invade upon people's lives through expanded surveillance and through the criminalization of just existing within the public sphere,” Mamdani said.
Except, nobody is calling the regular fare paying people criminals. Just the people who aren't paying the fare and breaking the rules.
Right, but you could theoretically say this about any software or technology (and this argument has very frequently in the past been used to argue against technologies like cars and airplanes). DNA sampling has "flaws" and "bugs" in it too that occasionally lead to false positives. Even police officers and lawyers falsely identify people occasionally. AI technology would just be another tool we collect evidence with for use in making inferences about the world. Imperfect technology never has been a blocker for using or improving that technology until it meets its purposes.
I feel like dismissals like this always lack imagination, you can't think of any uses for facial recognition surveillance that pegs you at specific locations at specific times? You can't imagine a single way how to monetize that data and use it against you or other law abiding citizens?
People said the same thing for social media data collection and now first degree price discrimination is getting more common. It's a total lie that it's only "people who are breaking the rules" suffer consequences with tracking and surveillance.
Just to note, everything is rosy right up until another MTA profit conversation and suddenly how can we monetize literally everything.
Also facial recognition is error prone enough that I'm sure a few people will get harassed by the system constantly flagging them. When they're not evading.
And evaders might wear a reflective hat or something making them invisible to the cameras.
> "you can't think of any uses for facial recognition surveillance that pegs you at specific locations at specific times? You can't imagine a single way how to monetize that data and use it against you or other law abiding citizens?"
Having imagination has nothing to do with staying on track and sticking to the conversation topic. Nothing you said has anything to do with criminalizing law abiding citizen for just existing. Privacy rights are a related but separate topic. We could talk about that separately if you want to. But I don't see how AI knowing where we are, or how AI enabling the monetization of our data has anything to do with criminalizing anything I do on a daily basis.
What facial recognition are you talking about when people jump over the turnstiles and walk into the open gates without any regard for police officers that stand nearby and just watch this happen? MTA is a joke. I keep wondering why am I even paying for the subway?
71 comments
[ 3.5 ms ] story [ 152 ms ] threadI did not sign up for anything special. I assume it’s from the real ID stuff.
Cats out of the bag, facial recognition is here to stay. Idk if I like it or not, but I can see how it will make my life easy.
When it works, I'm sure it's great. Now consider being a non-white person and it misidentifies you where you wind up being arrested for something you didn't do.
Until it works equally well for everyone, it should not be used for things other than processes of convenience. Even then, it should have a way to gracefully fall back to something else that does not prevent anything
[1] https://apnews.com/article/mistaken-arrests-facial-recogniti...
We've seen stories out of NYC where landlords have installed facial rec to access to tenants that didn't go so well. I can't remember if traditional key access was still available or not, but you could see where that would go bad if not.
If all people regardless of race, gender, or appearance were equally subject to the need to use the fallback then there would be less of an issue. But so far every commercial facial recognition program has demonstrated bias such that false positives and false negatives are much more common for PoC.
More recently people have discovered that if you pull the turnstile backwards a half turn you can squeeze by with very little effort.
This has been going on forever…
People need to get around. Taking mass transit is the best way for humanity for them to get there. Therefore, the fare should be zero.
Even if we feel a fare is needed, plenty of places operate with civilized proof-of-payment systems.
https://www.bloomberg.com/news/articles/2024-01-31/nyc-s-tra...
Also, turnstiles are not barbaric, what are you talking about?
Also since when is the rule that "the 'best' way has to be free"?
https://tfl.gov.uk/corporate/transparency/freedom-of-informa...
None of the articles I've seen ever seen to address why NYC has such a high rate of fare evasion. Is it just an ingrained "cultural" thing now? You could definitely jump over the TfL barriers if you wanted to, but no one does
I feel this is extraordinary dependent on how you measure "efficient" and how many false-positives or false-negatives are too many. (Also the fun problems where it's wrong way more often for certain groups of people.)
If the metric is solely "maximize the percentage of evaders who are punished", then the most "efficient" solution is to indiscriminately fine every single person. :p
If they actually arrest them, then there would be official numbers that align with the anecdata that anyone who has ridden that system is already familiar with.
The issue is not “facial recognition to prevent fare dodging”, it’s facial recognition to provide “reasonable suspicion” to warrant arrest and detainment of specific groups. NYC has a long documented history of creating laws solely for the purpose of racial discrimination - “stop and frisk” laws that functionally only applied to black residents for example, IIRC it was only the lack of probable cause that did anything to limit that. Systems like this launder the racism and provide “probable cause” for unconstitutional search and seizure, as well as false imprisonment.
It also means you get subject to the stigma of being stopped and questioned by police when entering a station while the wrong race - it’s obviously worse for black Americans whom the police routinely execute because they’re scared (specifically racist claims of super human strength, claims the victims are substantially larger than they are, etc), but these systems mean you’re more likely to be falsely accused if you’re any group other than white with european descent. So obviously black people are going to be victimized here, but Pacific Islanders, Hispanic, more or less any Asian country, etc are demonstrably subject to higher false positive rates, and there’s no penalty for the false accusations or detention.
Facial recognition software is equivalent to humans wearing a license plate. It's never been acceptable before and it's not acceptable now.
The MTA can't catch fare evaders because facial recognition is bad.
The NYC government can read peoples license plates and charge them arbitrary prices, partially, because the MTA is so broke.
So, fare evaders continue to get a free rides, in exchange for similarly Orwellian recording and logging of all vehicles entering and exiting certain parts of the city, which is effectively only one layer of abstraction different than scanning faces. Apples to pears, really.
It's what they do with the data in the long term that should be concerning in both cases.
You can’t see a distinction between the government’s ability to read a government-issued and totally optional identifier versus a… face?
Congestion pricing hits the lowest earning individual the highest. Congestion pricing should follow measurable improvements to public transit, focusing on the word "improvements".
A few rich people driving shitboxes around Midtown to avoid expensive fees sounds like a fine outcome to me.
This is true for anything that you buy. I’m not sure why being able to drive a car into the Manhattan CBD deserves special treatment. It also already has an excellent subway system that millions take daily.
You can’t means test virtually everything
Can you tell me more? In most places, that does seem obvious. In NYC, it seems less obvious to me. The friends of mine who complaining about congestion pricing rules are the upper/upper middle class folks who have not been on public transportation since Bloomberg’s first term. My friend who are lower income use public transportation everywhere.
https://www.youtube.com/watch?v=VOTaNK-8LYY&t=2s
Except, nobody is calling the regular fare paying people criminals. Just the people who aren't paying the fare and breaking the rules.
People said the same thing for social media data collection and now first degree price discrimination is getting more common. It's a total lie that it's only "people who are breaking the rules" suffer consequences with tracking and surveillance.
Also facial recognition is error prone enough that I'm sure a few people will get harassed by the system constantly flagging them. When they're not evading.
And evaders might wear a reflective hat or something making them invisible to the cameras.
Having imagination has nothing to do with staying on track and sticking to the conversation topic. Nothing you said has anything to do with criminalizing law abiding citizen for just existing. Privacy rights are a related but separate topic. We could talk about that separately if you want to. But I don't see how AI knowing where we are, or how AI enabling the monetization of our data has anything to do with criminalizing anything I do on a daily basis.