An AI dashcam app designed to rate every driver
spectrum.ieee.org
spectrum.ieee.org
And since its collecting ALL data, including visual and audio, this is a voluntary spy device. Imagine if the government demanded that data be handed over and starts using that to spy on people. IMO this is not a stretch of the imagination.
I can see one group of people having a legit beef with this though: car rental companies. One idiot behind the wheel, and their fleet car might as well go straight to scrap.
The opportunities for misinformation and asymmetricality are huge. Don't like someone? Clone their license plate for a day and drive like an ass. How could you ever prove your innocence?
Loan your car to a friend and they could tarnish your reputation. It would make it inhibitive to share vehicles.
This is technology that drives us farther apart as people in favor of a few insurance companies making even more profit than they do already. Think critically.
It would also give police even more excuses to profile innocent people.
If you want people to drive more safely, then think up ways to get more people to drive more safely. Don't expand the surveillance mechanism to every moving vehicle.
Unless, of course, they pay this nice fee to have their fleet cars whitelisted in the app, call it "bad driver protection money"!
All I can see this accomplishing is adding yet one more gadget app for people to be distracted by.
I tell this to anyone who rages: You drive and meet a LOT of people on your way. You don't make a lot of mistakes, but you make a few at least every drive.
Now given that you meet a lot of people you are most likely seeing a ton of mistakes by people. But individually they are not really making many mistakes.
Sure some people are horrible, but not all. And often the horrible ones are behind you.
And also, no matter how "safe" you are, it only takes 1 mistake to kill a person. Example, yesterday I was driving, a guy went nuts behind me and almost killed a motorcyclist. I saw that and it distracted me long enough to almost not notice the car that short-stopped in front of me. Literally 2 more seconds and it would have been a wreck. And I consider myself a safe driver, but it only takes one moment of distraction.
Basically what you're doing is playing Russian roulette, and you keep "winning", well yeah, that happens, right up until it doesn't.
Drive like a Google car. Stop being an idiot. Get a smiley face score.
This app and the related one about insurance liability (posted here maybe a week or two ago) would unfairly rate me, because I go against the risk model and win consistently. I also agree that there are MANY "safe" drivers whose indecision and timid driving leads to bad situations; merging at 45mph, slowing way down for ramps, stoping at yield signs, driving under the speed of traffic, braking unnecessarily in traffic, etc. However, since in my situation, the insurance company gets to charge me a lot more, they're incentivized against modeling drivers like me as less risky than someone who is involved in more accidents than me but drives well within the limits and doesn't understand what happens when they approach or exceed them.
I wouldn't expect that to be a yes/no thing. Shouldn't it be based on compared to the average driver, as determined by the same system?
People would love such systems (and I'm afraid our privacy is gone for good).
The real danger is the gradual erosion of individual liberties through the automation, integration, and interconnection of many small, separate record-keeping systems, each of which alone may seem innocuous, even benevolent, and wholly justifiable.
The whole report is public, and it specifically addresses insurance in this context: https://aspe.hhs.gov/report/personal-privacy-information-soc...
looking -> recording -> sharing
These are all different things and our behaviour will and should change at each step.
No one can be drowned by a raindrop. A flood is another matter.
If I'm not that one person I'd have a problem with that.
> why suddenly connecting these dashcams to a network that shares information is a problem?
Switching from dash cams to police helps a bit. I think most people would be okay with one (or a handful) of police officers randomly walking through their neighborhood. Most locales would say it's a good thing.
Now take that to the extreme and have an officer every 10 steps, looking around in every direction, and writing down everything that's happening. Still a good idea?
Probably that bit
What if he was driving like an asshole because he was speeding to the hospital or similar reasons?
A real-life example: my friend was dating a woman whose brother got in an accident. After receiving painkillers, he became addicted to them (an old story at this point). As he weened himself off of them, taking responsibility, my friend would drive him to get methadone a few times a week. A system that uses information from this one would label my friend an addict because he was so often seen driving to or parked at the rehab clinic. When in fact, my friend was a charitable, stand-up guy helping out a soon-to-be family member recover and become a stronger, productive member of society.
On a more broad note, I wonder how useful it really is to know that a specific vehicle acted recklessly sometime in the past. Seeing someone behaving poorly is good reason to avoid them on the roadway. However I also imagine a scenario where one driver did something to warrant an alert, and some time later every car on the roadway is trying to avoid that vehicle. Is there really any danger from that vehicle in the current situation?
What I would like to see is a map showing where the most incidents occur, and what type of incidents occur there. I assume most drivers genuinely set out intending to behave well, but do not for various reasons. Identifying locations where incidents happen would give civil engineers some valuable data about how their designs impact behavior, and would give police valuable information about where their presence would be useful.
Have you ever driven full speed through a red light? Sat at a complete stop of a freeway in an attempt to merge in to traffic? Taken a U-turn on a freeway? There are things that don't cause an accident, but once a driver does, they should probably not be allowed behind the wheel for at least a year, some things possibly even a lifetime ban or some additional evaluation.
There is an interesting dichotomy because smart phone usage has really caused a big problem with distracted driving. Even Geico experienced an increase in claims last year, and I suspect this was why. When I watch traffic sometimes over 50% of drivers are looking at a phone screen rather than the road. At the same time, mobile sensors are enabling something which will end with roads being extremely safe and killing very small numbers of humans a year.
There is a lot of talk about privacy and freedom on the internet. Privacy and freedom on the road is dead in all dimensions. You won't be anonymous, you won't be able to take go somewhere when the police/authorities are looking for you, there may not even be private ownership of any vehicles in the US in 10-15 years.
What we should be interested in is the top one percent.
If you're a "good driver," as defined by a lack of tickets and accidents, you get discounts. There's really no way to tell if you're _actually_ a "good driver" because "good driver" is a hidden variable, and the only observables from an insurance standpoint are tickets or accidents. I know horrible drivers who never have accidents and avoid tickets via luck or skill. Conversely, you can be a good driver who's unlucky enough to have multiple accidents in a short period of time. Outcomes = baseline truth + random noise.
Most corporate data grabs bad news, but I'm ok with it when it comes to pricing insurance (outside of health). If I run a business that has rock solid, well enforced policies in place to mitigate risk, I don't want the insurance company's priors over my company's sector driving the premium. I only want to pay for coverage on idiosyncratic risk - the risk that remains after addressing everything that I can. Same goes for driving habits. Things like health are almost entirely idiosyncratic risk (ergo, let's agree that doing this for health is a bad thing), but when that's not the case pricing policies using true states (the baseline) and not observables (baseline + noise) puts you as the policy buyer more in control, not less.
That said, I have major concerns over the implementation. A breach is all but inevitable, so there needs to be a means of updating models online, and rules against data retention. It's also bad news if flawed models or corner cases incorrectly price someone out of a policy and there's no sanity check. Unfortunately, that's not too different from the actuarial status quo.
Actually it's quite rare for insurance companies to record an underwriting profit, I believe. Their profit generally comes from investing the pool of premiums. If you invested your premium instead of buying insurance however, then I suppose the expected return would have to be better.
This would only be true if the value of money was linear.
The value of money is not linear.
A 10% chance of a 1000$ loss can be far more harmful than a 100% chance of a 100$ loss, especially at low income.
Most people on HN knew about expected value but they don't know too much about expected utility. I suggested reading about it and St. Petersburg paradox
Absolutely. I'm not one to normally fall victim to the mindless "think of the children", but Amber Alerts are rare and specific enough that they seem perfectly useful and legitimate usage of this tech. (I find most of the rest of the tech between useless and invasive, but flagging Amber Alert tags I'd support.)
Amber Alerts for my area come to my cell phone and I can't recall getting more than one per year. Looking up stats for 2014, they ranged from 9-21 per month (nationwide). I'm pretty comfortable that surveillance over-reaches will be a long time coming to Amber Alerts (if ever).
In my area, until I turned them off, I received at least 2 alerts per week and the vast majority of these were domestic-related (as in, one true parent of the child had the child when they were not legally allotted to - and the other parent reported it).
From my observation Amber Alerts have always been domestic squabbles. Before I turned them off on every device in the house, I don't recall a one that wasn't a divorce dispute. In other words, not a one was a reason to make every phone in the office make noise. In other other words, we collectively found out that all of these "abductions" wouldn't fit within the definition that most of us use for the word.
As is typical today, it uploads all the data to a server and the service operator keeps all the rights to the data. Sigh.
Using it to pass around hard braking info is not useful, though. That's what radars are good for. They'll always see the car ahead, whether it's equipped or not. The widespread use of radar-controlled auto-braking is the next big thing in auto safety. BMW, Mercedes, etc. already have it, so it's a proven technology. Not car to car communications, which don't solve any non-advertising related problem.
Can anyone speak to this? Am I observing this on some kind of bias?
Also, Stephen Hawking himself has recently participated in a boycott of Israel: [1]
[1] http://www.cbc.ca/news/technology/is-stephen-hawking-s-acade...
There'd have to be a very big reward for me voluntarily becoming Big Brother, and especially for tattling on myself.
I am very interested in what can be done, with such data. I am currently working with long duration (>20 minutes) dashcam videos uploaded by users to youtube.
We can't have companies violating privacy like this. Only the government can do that.
Why would the government want to spend the money and PR capital developing and rolling that out, when they can have a company do it for free, take the heat, chest beat about private enterprise, and then secretly NSL subpoena or crack it and distribute the intelligence to every interested agency, police department, and eventually favored corporate interest (using parallel construction as needed)?
And while your insurance cost might also go up if it deems you to be a reckless driver that can also have a good impact because it can force you to drive more safely in order to bring your insurance costs back down.
If they can incorporate the app with some instructional service like a weekly summary of all the misstakes you make and give you say 5 tips to drive better it can actually have quite a positive effect on road safety.
And while it's a bit "creepy" I see this to be pretty much similar to insurance companies "paying you" to quit smoking even sending you to "free" quitting smoking seminars/workshops or "paying you" to go to the Gym often even with a free yearly subscription.
Those things cost pennies to the insurance companies while saving them a lot of money in the long run, only a handful of less lung cancer cases or heart attacks a year can cover the cost of those programs for 1000's if not 10,000's of policies.
*By "their videos" I mean videos from users using the application that get uploaded to their servers and posted to their YouTube channel.
Probably would get a demerit from an AI for that maneuver.
I wonder if they'll be able to tell when someone else is driving your car from the "fingerprint" of their driving style.
Pretty interesting stuff
My roommate in grad school developed a somewhat related idea but for identifying someone's "typing cadence". He used it for an additional security factor in the way that a person types their passwords in a browser.
Google has one, a few banks use it to identify potential fraud cases appearntly even simple things about how you navigate a webpage can produce a fairly unique fingerprint even across different devices.
For those who aren't cyclists, there are already informal networks of people who share helmet-cam videos of bad driving behaviour, with people who attempt to read license plates etc
As a driver, I have been at the receiving end of abusive gestures and language even when following the law. In one case, I yielded to a bicyclist who was in blatant violation of the law[0], because failure to do so would have endangered him. He showed me his middle finger. Another was angry that I was in the bicycle lane making a right-hand turn, as required by law and common sense.
As a cyclist, I have given up trying to educate my fellow cyclists, and generally won't ride with others anymore.
[0]I was already turning through the intersection after having made a stop. The bicyclist arrived in the middle of my turn and entered the intersection without stopping for the stop sign.
Edit: Formatting.
I don't know about the bay area, but in my city you can't really group ride, bike roads are single lanes. Yet people ride side by side all the time.
Personally, I found the best way to change the culture is to show an example. There is clearly an ongoing bicycle boom, and it shows many just follow the customs e.g. dashing through reds is acceptable (which is debatable) but it quickly turns out stopping for 20 seconds is not that big of a problem. It is quite interesting though, the same people would never do that in a car.
I recently watched a video compilation put together by a truck driver who intended to show the dangerous behavior of passenger vehicles. It showed what it intended, but it also showed dangerous truck driver behavior. When this behavior was pointed out, the truck drivers in the comment section came up with many excuses for why this behavior was acceptable. One actually argued that in one of the scenarios the truck driver should not have had to yield (despite being required to do so by law) because accelerating back up to speed was expensive in terms of fuel.
So there was proof positive that this truck driver not only violated the law, but created an unsafe situation in a video specifically edited to show it was not truck drivers but passenger vehicles which caused unsafe situations, and this group of truck drivers defended the behavior.
In that case, the truck driver was ignorant and stupid. His behavior wasn't changed. And I see that a lot in YouTube videos about drivers of vehicle X complaining about drivers of vehicle Y. Anyone pointing out mistakes made by the driver of vehicle X is shouted down, nonsense about what is legal is posted (most of it wrong or incomplete), and everyone who drives a vehicle X shares their own (possibly apocryphal) stories about people driving vehicle Y.
So no, even with camera, I don't care much about the opinions of cyclists on drivers. Cameras can be good, but they can also be bad or indifferent, and nine times out of ten, I've got questions even after seeing the video.
1) Only have the information for licence plate codes of people who have opted in be stored in a way associated with the specific licence plate code.
2) Have the system encourage opting in as part of setting up the system in one's own car.
3) Have an id which is used for all cars which have not opted in, so that cars which have not opted in have a collective rating. Possibly split this up by general location, and maybe time of day.
4) Because most of the cars in a given area are there often (or, at least something similar to this is true) , it may be worthwhile for, for the cars that do not opt in, to still record behavior of those cars, but to not send it to a central location, only keeping it within one's own car. Cars which have not been seen in a long time can be automatically removed from the local storage. This should not significantly harm privacy, because it is basically the same as remembering that you saw that car before and that it tended to be driven poorly.
5) Possibly (this might always be pointless/counterproductive. I'm not sure.) , in some cases only store a hash of the license plate codes, instead of the license plate code itself?
A) I think this variation should be sufficient to protect privacy, because it only does widespread collection about identifiable individuals if they they consent to it. The information it collects about individuals that did not specifically (and presumably intentionally) express consent is only collected in aggregate about individuals in a geographic region, which does not personally identify anyone, or only collected by individuals and not shared, which is not much more of a privacy violation than an individual with excellent memory looking around.
B) So, while this variation might be less effective, I think it should be sufficient to protect privacy.
This tech would be susceptible in a similar manner - stolen vehicle goes out and drives badly etc. etc. The only identifier would be the licence plate. How about using their phone account as the identifier rather than the vehicle?
Anyway sounds like an idea ripe for misuse and "feature creep".
Also think of this way: if you both tailgate, and you have to brake, you will be hitting both the one in front of you, and getting hit in the back. If you don't tailgate and only get tailgated, you will only get hit in the back. (And have a way better defence for your insurance company.)
(PS I'm going to assume this is tailgating on the highway ?)
Google sees the car in front is a Tesla: 1 star :)
Both Yelp and Travelocity turned from awesome to pretty much useless within a few years. Reviews and ratings can be bought. Employees are forced to leave positive reviews.
It's also true with credit cards. If you don't understand merchant fees and maximizing your rewards from purchases you'd otherwise make while simultaneously bolstering your credit score, you're leaving money on the table.
Humans will hopefully look back in a few hundred years and realize how futile and terrible of an idea it was to attempt to gamify every aspect of our lives so some company or another can extract a profit from the unsavvy, but since there is so much "value" to be "created" right now, I doubt that it'll happen any time soon under our current system of governance and beliefs about property, taxes, "free" markets, etc.
Think about instant posts of driving achievements to Facebook, instagram, twitter.