> "Insider reported in February that Amazon was equipping all delivery vehicles with an AI camera system called Driveri, which is manufactured by a company called Netradyne.
> The cameras are always on and scan drivers' body language and the speed of the vehicle, detect if a driver is wearing a seatbelt, and even measure drowsiness.
> The system then uses "automated verbal alerts" to tell drivers if a violation has been detected."
> [...]
> Amazon told Insider in February driver footage was not automatically available to Amazon and that the "live feed" was triggered only after the detection of a safety or policy violation. Amazon did not immediately respond to a request for comment on this story.
Wow this sounds like a dystopian science fiction movie.
It's scary to notice that Amazon PR team gets away with just saying "we promise it's not automatically available, and that we don't abuse it by surveilling 24/7". All this just reminds me of the creation and abuse of Uber's god mode, Snowden's revelations of NSA's secret dragnet surveillance, etc.
Things like this are only appropriate to use when the code for the system is 100% open source. It's anti-democratic when a system of rules that can instantly upload a live feed, which is "triggered after the detection of a safety or policy violation", is not visible to a participant in the system. It's tyrannical and terrifying. Systems like this just end up gaslighting workers and invasively monitoring them. It uploads their private information (it monitors their 'body language' - whatever that means) to a black-box vault, to be used for further exploitation by capital (and fed into the capitalist, racist and sexist Silicon Valley AI monster).
"Moral Statute Machine:
Your repeated violation of the Verbal Morality Statute has caused me to notify the San Angeles Police Department. Please remain where you are for your reprimand."
These also sound like the excuses you come up with to install cameras in every car when you can't say the real reason because of bad PR.
This approach works really well, because you can just accuse people who are against your proposals as being 'against driver safety'.
It's similar to every other move in increasing surveillance in that you can come up with a 'PR-friendly' reason for the surveillance that people find tough to argue against to placate the masses. "Oh, you are wondering why we are spying on everyone... uhh... it's to protect the children! It's to protect the children! And terrorists! We need to stop the terorrists! And the people who are against these proposals? Well they must be people who are pedophile-enablers and terrorist sympathisers!"
Shrinkage. They're trying to point the finger at who's losing their packages. If some sorter scans a package and their tracking system says its in a certain tote number they're going to assume its there even when, in point of fact, it's not. So when a driver reports a package missing... they want to know he didn't just shove it in his bag.
Note that this is readily solvable by having DSP and Amazon Logistics drivers do what Amazon Flex drivers already do: scan every single package individually. If a package comes up missing then flag an Amazon rep to inspect the bag its supposed to be in, verify its missing, snap a quick pic of the tracking#, and then they can take it off your route while they try to find it (and ultimately wind up marking it missing later). That, of course, would mean drivers take more time loading... so of course they'll never do that. They'll just install cameras.
To an extent, they already are. Most new class 8 trucks in North America are so connected to the rest of the fleet it's scary. A lot of trucks can be re-programmed by the company while it's out on the road. Sending the status of the seatbelt, coupled with lane monitoring systems that can phone-home as a violation if done x number of times in y <time frame>, can easily implement the same goals at virtually zero cost.
Systems like this are for what you can't use existing vehicle sensors to do - detect if the driver is eating, drinking, using a cell phone (believe it or not some mega-carriers even have no-hands-free policies), reaching to adjust controls or grab something, talking to a passenger(s), doing drugs or drinking, the list goes on. They're meant to put as much liability and proof-of-fault as possible (or not) on the driver, and the whole driver-facing camera thing in general can drive down carrier insurance rates. "Oh the driver was reaching for his coffee and couldn't brake in time while someone cut them off to make their exit?" can sound great to some people to have on video, but "Driver was fired for eating a Big Mac while driving" or even "driver was caught masturbating in company truck" are very real things that have happened to people I personally know.
Luckily, there has been precedent set here in Canada that has banned driver-facing cameras for privacy-related reasons. Hopefully it stays that way, and the US follows soon. Whatever your views on how much safer it may make the roads, it's a gross invasion of privacy. I liken it to having your boss install a webcam in your office that's watching you 24/7, I'd bet a large majority of people would not put up with that.
Without visual confirmation, the driver could either buckle the seatbelt under themselves (buckle over the empty seat and then sit on it) or have a buckle (yes, just a buckle) to trick the sensor into thinking all is well. I’ve seen taxi drivers do this to avoid having to actually buckle up; I guess it’s uncomfortable to them?
I wonder if wearing a shirt with an image of a seatbelt would be enough to trick the camera or AI :)
Buckling the seatbelt and sitting on it (implying that they're going to just buckle once and never bother with it) is easily detected via vehicle telemetry. They track every time you buckle and unbuckle the belt. Anyone who isn't buckling and unbuckling at the expected rate is flagged. Occasionally this results in the discovery of a faulty sensor which gets replaced, but nine times out of ten it results in the discovery of someone violating company policy and thus incurring an infraction. They don't need a camera for this.
The car itself tells you when it expects you to buckle up by beeping annoyingly, so it sounds like beating this method just requires the little bit of discipline to unbuckle your “phantom buckle” every time you get out of the car.
Honestly that doesn't even really save you any time. Neither does leaving the side loading door open constantly (though I've witnessed drivers from other DSPs doing that, too). That's another thing that vehicle telemetry can already tell them, so this is still simply unnecessary.
Oh no contest from me - I’m just telling you what I’ve seen in the wild :) also I don’t think it’s about saving time, but about the seatbelt being uncomfortable over long periods (observed this with taxi drivers who are behind the wheel a lot - another thing I’ve seen is, they buckle up when entering the highway, unbuckle when exiting into a residential neighbourhood - police presence and fines for not having the seat belt on are tougher on the highway. At this point they are actually wasting time vs. just leaving the seat belt on all the time, so it must be because of the discomfort... I guess)
"Netradyne provides cutting-edge technologies in AI, ML and Edge Computing to help reduce accidents by creating a new safe driving standard for commercial vehicles. Our industry solutions reduce driving incidents and protect against false claims"
> Right now it's just a way to gaslight workers and invasively monitor them and upload their information to a black-box vault, for further exploitation by capital (to feed into the racist AI monster).
Just wait until they deploy something like this for the warehouse workers and they start getting flagged every time they scratch their nose.
I don't know how it is in the US but I always get the feeling the Amazon driver uses his own van or car to deliver the parcels. Wouldn't even think about having something like this installed in my own car.
I don't agree with such a system but if you use Amazon's van/car there is less opportunity to be against it.
In the UK at least there is a mix of the following:
* Independent couriers (via a delivery app similar to Uber).
* Small business owners that will have a fleet of c10 vans / drivers that contract for Amazon (via a specific Amazon programme to start up your own small van-delivery business for delivering Amazon parcels).
* Amazon directly-employed van drivers.
I think independent couriers are charged a cost per parcel delivered anyway, so it doesn't make sense to monitor those any more than you need to. This potentially could then apply to points 2 & 3 above depending on how far Amazon wants to go with it.
I actually didn’t give that any thought until your comment, but now that I think about it, I haven’t seen a single Amazon package being delivered to our section of neighborhood (east coast US) in a privately owned vehicle for maybe a year or more. It used to be fairly common, but I believe I’ve seen only grey-blue Amazon vans for quite some time now.
In the UK that would tip the driver over IR35 and they could have a case for employment tribunal, but in about 2 weeks changes to that are coming, and it will be legal for Amazon to do it while not being liable for employment laws.
I'm not sure open sourcing entirely solves the problem, that doesn't show how the company has the software configured. I'd suggest that you need rigorous and continuous independent auditing with results published. But where do you get that these days?
> The cameras are always on and scan drivers' body language and the speed of the vehicle, detect if a driver is wearing a seatbelt, and even measure drowsiness.
> The system then uses "automated verbal alerts" to tell drivers if a violation has been detected."
> [...]
> Amazon told Insider in February driver footage was not automatically available to Amazon and that the "live feed" was triggered only after the detection of a safety or policy violation. Amazon did not immediately respond to a request for comment on this story.
Wow this sounds like a dystopian science fiction movie.
It's scary to notice that Amazon PR team gets away with just saying "we promise it's not automatically available, and that we don't abuse it by surveilling 24/7". All this just reminds me of the creation and abuse of Uber's god mode, Snowden's revelations of NSA's secret dragnet surveillance, etc.
Things like this are only appropriate to use when the code for the system is 100% open source. It's anti-democratic when a system of rules that can instantly upload a live feed, which is "triggered after the detection of a safety or policy violation", is not visible to a participant in the system. It's tyrannical and terrifying. Systems like this just end up gaslighting workers and invasively monitoring them. It uploads their private information (it monitors their 'body language' - whatever that means) to a black-box vault, to be used for further exploitation by capital (and fed into the capitalist, racist and sexist Silicon Valley AI monster).