This thread is deliberately wider in scope than just AI drivers.

However one point about them occurs to me

The most common driving offence is speeding.
Does any proposed AI driver always observe speed limits ?
If not why not and who is responsible?

On 12/7/2025 at 12:00 PM, studiot said:

This thread is deliberately wider in scope than just AI drivers.

But the scope does focus on this, bc it's the first interaction that can be measured in a meaningful way, the greater 'we', are happy to accept AI is 'smarter than the average bear', if your a "professional chess player", but I'm obviously smarter than that bc they're just nerds.

On 12/7/2025 at 12:00 PM, studiot said:

The most common driving offence is speeding.
Does any proposed AI driver always observe speed limits ?

The AI's in my mates car always admonish him whenever he, or other's, breaks the rules... It's bloody annoying, but I've found a work-around... 😉

If I bounce on the seat a couple of times, it thinks I'm wearing a seat belt...

On 8/25/2025 at 1:39 PM, studiot said:

Apparantly the AI will monitor for such things as not wearing seat belts, talking on the phone whilst sriving, eating whilst driving, children and animals incorrectly secured in the cabin and so on.

I don't think that in the near future it will be able to detect the examples you gave. Detecting seat belts when someone is wearing black clothes and/or has tinted windows? That's not very realistic. *) Detecting children or dogs in the back seat?! The rear seats are not visible at all to such cameras.

Detecting license plates is difficult enough in itself. The faster someone is driving, the more blurred the image is. At 70 km/h, a vehicle travels 19.4 meters in one second. At 30/60 fps, this gives ~ 65 cm/frame and ~ 32 cm/frame, respectively.

Your phone has a sports mode that allows you to take better quality photos by shortening the exposure time. But this, in turn, means that nighttime, cloudy, and rainy conditions will be recorded less well. This sport mode only reduces blurring, it does not eliminate it completely. It registers fewer photons. In order not to lose quality, it must be took in good lighting (lots of photons).

*) ChatGPT calculated that 1 pixel of a 4K camera (3840 x 2160 pixels) corresponds to 3 cm when the camera is 100 m away from the car.

I think the whole process will look completely different. AI will analyze the car at point A (first camera) and look for it at point B (another camera). The distance between A and B is constant and known, so we just divide by the number of seconds and check if it exceeds the average. So, for example, our 19.4 m/s for 70 km/h limit. If so, it will search for that car on the exits from highway, comparing only the appearance of the vehicles. And there, at low speed, it will get a clear image of the license plates.

Human assistance will still be needed to check if plates were good enough and cars were not mixed.

Detecting someone who is speeding (or riding with someone who is speeding) can be done by analyzing the login logs to GSM network base stations. If someone does this regularly, they will often be at the very top of such a database, so all you have to do is wait for them to drive off the highway.

You shouldn't go overboard with enforcement here. Sometimes speeding up and exceeding the limit is necessary to avoid an accident.

ps. Now imagine how absurd this is: cameras to catch drivers and give them tickets are everywhere, people have them in their cars, especially in Western countries, but planes don't have video recorders to capture what pilots are doing, even though the lives of hundreds of passengers depend on their mistakes.. When pilots report an emergency, all data from their cockpit could be sent to headquarters in real time, and a whole team of people could analyze it even before the crash. Depending on the type of fault, this drop from 10 km to 0 can last up to 15-20 minutes after engine failure.

6 hours ago, Sensei said:

Detecting license plates is difficult enough in itself. The faster someone is driving, the more blurred the image is. At 70 km/h, a vehicle travels 19.4 meters in one second. At 30/60 fps, this gives ~ 65 cm/frame and ~ 32 cm/frame, respectively.

Not sure how this matters. Plates are on the front and/or back of cars, so the cameras tasked with this effort are pointed to mostly align with the direction of motion, not perpendicular to it. The pictures from the toll cameras I got last year were quite clear, and that was at ~100 km/hr. Getting a meter closer to the camera did not cause much blurring.

Detecting someone who is speeding (or riding with someone who is speeding) can be done by analyzing the login logs to GSM network base stations

You could use radar, like they’ve been doing for years.

Speed cameras are 1/1000sec or better. Some new photo cameras can do 1/32000sec. The cameras calculate exposure parameters based on vehicle speed and lighting conditions. I think radar data is integrated into the exposure formula.

On 12/6/2025 at 6:03 PM, swansont said:

But that won't matter; the liability lies with whoever is at fault. So I think that we won’t get true autonomous vehicles until the companies accept that liability, or con the customers into accepting it. But to get to the customers to accept the computer being better than the average driver isn’t enough, because most people think they are good drivers. The computer has to be better than people think they are.

And that includes accident avoidance that I mentioned earlier - I think most people won’t accept getting into an accident even if it’s the other driver’s fault, owing to injury (money compensation vs chronic pain/permanent disability) and just the hassle if getting a car fixed, even at minimal monetary cost.

ETA- One big issue is/is going to be convincing people they’re at fault when a human and computer vehicle get into an accident. Another is issues with accidents involving pedestrians

I don't disagree with any of this. One thing is, most new vehicles have cameras so I don't see why automated vehicles shouldn't have a recorded 360 view to determine such things.

2 hours ago, npts2020 said:

I don't disagree with any of this. One thing is, most new vehicles have cameras so I don't see why automated vehicles shouldn't have a recorded 360 view to determine such things.

There might be pushback against cars actually recording 360 degrees of their surroundings, by our techbro wanna-be overlords.

On 12/9/2025 at 10:10 PM, swansont said:

There might be pushback against cars actually recording 360 degrees of their surroundings, by our techbro wanna-be overlords.

Possibly, but there doesn't seem to be much uproar about cameras being on virtually every block of US cities or cell phones being able to track everywhere one has been with it.

2 hours ago, npts2020 said:

Possibly, but there doesn't seem to be much uproar about cameras being on virtually every block of US cities or cell phones being able to track everywhere one has been with it.

It’s one thing for the government to monitor specific areas (public areas), and at least in the US you need a warrant to get tracking info from a cell phone. If it was discovered that this was happening without one, there would likely be an uproar.

Like what has happened with facial recognition

https://www.nbcnews.com/tech/security/facial-recognition-technology-use-stadiums-us-sparks-protests-rcna167410

“A group of eight people with the nonprofit advocacy group Fight for the Future and the New York-based nonprofit Surveillance Technology Oversight Project (STOP) protested against facial recognition scanning and other biometric data collection outside a Mets game, trying to raise awareness among attendees about what was happening to some ticket-holders when they stepped into the stadium.”

2 hours ago, swansont said:

and at least in the US you need a warrant to get tracking info from a cell phone. If it was discovered that this was happening without one, there would likely be an uproar.

Don't you have apps to monitor the whereabouts of pets (dogs) and other apps to monitor your children ?

9 minutes ago, studiot said:

Don't you have apps to monitor the whereabouts of pets (dogs) and other apps to monitor your children ?

I believe so. But it’s not the government doing it, and if the techbros were accessing the data there would be backlash. Like Amazon being sued over Alexa violating users’ privacy (recording and storing without knowledge or consent)

https://topclassactions.com/lawsuit-settlements/privacy/amazon-alexa-class-action-alleging-privacy-violations-moves-forward/

Similar apps and devices are also used to monitor (senile) elderly with the habit of wandering off

13 hours ago, studiot said:

Similar apps and devices are also used to monitor (senile) elderly with the habit of wandering off

The same apps can be used to monitor children, for their own good, very nebulous in legal term's and liable to mission creep, even in a moderately strong right/left wing society. And it takes a very, very obvious corrupt miss-use for a government legislate against the insidious creep.

Edited December 13, 2025Dec 13 by dimreepr

How long until robot vacuum companies get sued for collecting data, and phoning home without consent?

https://www.zmescience.com/science/smart-vacuum-companies-are-monetizing-maps-of-your-house/

“Those cute little robot vacuums buzzing through your home may be doing a lot more than dusting the floor. They could be learning details about your house and making money off of that.”

I should remember to tell our damp mop and brooms how much we love them. Don't want them rising up like in that Fantasia scene.

36 minutes ago, TheVat said:

I should remember to tell our damp mop and brooms how much we love them. Don't want them rising up like in that Fantasia scene.

Don’t axe, don’t tell.

10 hours ago, swansont said:

How long until robot vacuum companies get sued for collecting data, and phoning home without consent?

https://www.zmescience.com/science/smart-vacuum-companies-are-monetizing-maps-of-your-house/

“Those cute little robot vacuums buzzing through your home may be doing a lot more than dusting the floor. They could be learning details about your house and making money off of that.”

iPhone and Apple do the same thing, and even a thousand times more, and yet you don't mind buying and supporting them every year... ;)

If you can control an IoT device by voice, it must have its speaker turned on non-stop to detect that you are talking to it. The same applies to phones that have a voice-controlled assistant. So if you agree to send debugging data (you have such a clause during Windows installation, for example), you also agree to send your voice if it is needed. Every computer and device has a unique serial number, and computers have several or a dozen such serial numbers for different parts. Each of them can be used to identify someone. You don't know who stole your phone, the police don't know and don't care who stole your phone, but Google knows who stole your phone... ;) as soon as they try to create a new Google account or log into an existing account (even after resetting the data on the phone) (without it Google Play Store won't work)..

If someone is “synchronizing data” between two devices because they bought a new phone, how do you imagine this process? It could be sending everything from phone A to the cloud, and then from the cloud to phone B. So you gave all your sex tapes and nudes and all the secrets of your life to Apple/Google...

The most dangerous thing is that if these devices need to download/upload data from the Internet, you give them access to your internal LAN, which in most cases means your Wi-Fi password. This is 100 times more controversial than the fact that this little robot drives around your carpet and knows the square footage of your room.

Hiding the fact that someone is viewing your Wi-Fi password would be child's play under the guise of IoT device debugging data.

BTW, I heard that one of the most famous cybersecurity here experts got hacked. He revealed it himself to show how easy it is to do. Fortunately for him, the hacker was a pentester and told him himself that he had screwed up. The guy made a website using AI, some kind of “AI Cursor.” 18k+ lines of code generated. Web hosting project. The AI set up the Apache server itself. People told him that directory listing had been forgotten to be disabled, and what were the consequences? Someone (pentester) discovered that one of the directories was .git. And it had the entire source code of the entire website project. The pentester downloaded 2700+ files from this project. He read and analyzed the source code. And he created his own 70 GB VPS, just like that, with access to all the server's databases.

Edited December 15, 2025Dec 15 by Sensei

6 hours ago, Sensei said:

iPhone and Apple do the same thing, and even a thousand times more, and yet you don't mind buying and supporting them every year... ;)

They’re mapping my house or apartment and sending the data somewhere? Evidence?

6 hours ago, Sensei said:

If you can control an IoT device by voice, it must have its speaker turned on non-stop to detect that you are talking to it. The same applies to phones that have a voice-controlled assistant.

That was addressed a few posts back.

6 hours ago, Sensei said:

So if you agree to send debugging data (you have such a clause during Windows installation, for example), you also agree to send your voice if it is needed. Every computer and device has a unique serial number, and computers have several or a dozen such serial numbers for different parts. Each of them can be used to identify someone. You don't know who stole your phone, the police don't know and don't care who stole your phone, but Google knows who stole your phone... ;) as soon as they try to create a new Google account or log into an existing account (even after resetting the data on the phone) (without it Google Play Store won't work)..

If someone is “synchronizing data” between two devices because they bought a new phone, how do you imagine this process? It could be sending everything from phone A to the cloud, and then from the cloud to phone B. So you gave all your sex tapes and nudes and all the secrets of your life to Apple/Google...

You're missing the point. It’s sending private information, that isn’t necessary for the operation of the device.

People aren’t upset that you’re syncing data that you told it to. Or that Alexa is listening, it’s that someone else is.

On 8/25/2025 at 7:39 AM, studiot said:

Considering all the recent discussions about AI lying to us to satisfy its programming, how should we consider evidence of wrongdoing provided by AI ?

AI does not actually lie, but neither does it tell the truth. We tend to associate intelligence with being smart, honest, truthful, etc., but it is not. Measured intelligence is simply a speed test in something's/someone's ability to assimilate information. So AI (artificial intelligence) is all about statistics and has nothing to do with smart, honest, or truthful. Courts, on the other hand, are all about truth, so what happens when you mix these two ideas -- intelligence and truth?

Consider that forty or so years ago, the State of Michigan (USA) was all about trying to convince it's people to use seat belts, so there were signs everywhere stating, "Seat Belts Save Lives". It was on commercials on television, on radio, on billboard signs, everywhere -- then it stopped. I was surprised that the campaign stopped until I learned that the Michigan Supreme Court heard a case regarding seat belts and concluded that they did not in truth save lives and sometimes they killed people. In light of this new information, I expected the seat belt law to be retracted -- it was not. We still have to wear seatbelts.

What I did not know, but learned later, was that seatbelts do not save lives, but they do save money. As far as injuries go, it progresses from minor injury, to major injury, to death, but as far as expense goes, it progresses from minor injury, to death, to major injury. Apparently a major injury is much more expensive than a death, and seatbelts prevent major injury -- wearing a seatbelt will protect you from major injury, or it will kill you. Much less expensive -- truth.

Something to think about.

Gee

3 hours ago, Gees said:

Consider that forty or so years ago, the State of Michigan (USA) was all about trying to convince it's people to use seat belts, so there were signs everywhere stating, "Seat Belts Save Lives". It was on commercials on television, on radio, on billboard signs, everywhere -- then it stopped. I was surprised that the campaign stopped until I learned that the Michigan Supreme Court heard a case regarding seat belts and concluded that they did not in truth save lives and sometimes they killed people.

Citation needed for this court case

18 hours ago, swansont said:

They’re mapping my house or apartment and sending the data somewhere? Evidence?

Seriously?

https://edition.cnn.com/2019/08/19/tech/siri-alexa-people-listening

If you're walking down the street and suddenly an advertisement for a store you're passing pops up, it's no coincidence—your phone sends its location (not via GPS, but GSM), and advertisements are selected based on that. When you log into your Google account, you will find a timeline of places you have visited in recent days and months.

https://support.google.com/accounts/answer/14200149?hl=en

Knowing where you are (where you live), getting information about the layout of the rooms in your apartment is a piece of cake.

BTW, several hundred readings from GPS/GSM and the size and layout of the rooms are known.

18 hours ago, swansont said:

You're missing the point. It’s sending private information, that isn’t necessary for the operation of the device.

This is not private data, just diagnostic data. If you had read the terms of service, it would have mentioned this and you would have agreed to it yourself.

And I disagree, they are necessary. They can be used to e.g. calculate how quickly a given model breaks down, i.e., after how many kilometers there are breakdowns.

It does not transmit GPS/GSM data, only acceleration data. Acceleration can be used to calculate speed, position, and distance traveled.

If you bought your laptop in a store, did it have Windows pre-installed? Did you reinstall it right after purchase? During installation, Windows asks if you want to send diagnostic data (which could mean anything). I bet you didn't reinstall it five seconds after getting home..

i.e., the person who installed this pre-installed Windows or macOS, Android or iOS agreed to the terms of service on your behalf. ;)

Last month, I saw something strange (at least to me): a guy (who knows a lot about cybersecurity—he even has a special card without which his laptop cannot be started) received a package from China with the best cell phone in the world in terms of specifications (not an iPhone), and he opens it, and the whole system is ready to go... I tell him, “Format the data! Reinstall!”... but he says no, no... What a complete idiot... even a regular DHL courier could have uploaded malicious software to it on the way to us...

Edited December 16, 2025Dec 16 by Sensei

On 12/15/2025 at 10:43 PM, swansont said:

Citation needed for this court case

I am sorry swansont. If you had asked for a citation twenty-some years ago, I could have accommodated you, you know, before MS stole so many of my abilities including the ability to work, where I would have had access to a law library -- the kind with books -- but all I have now is the internet and have no idea of how to look things up. I tried, really I did. I was surprised to find that there are still a lot of cases regarding seat belts right up to 2025, but I don't know how to narrow the search.

The original case that I had mentioned was not about whether or not we should use seat belts -- it was already law that we were required to use them -- it was about advertising the seat belts. Stating that "seat belts save lives" was inaccurate, a lie, because seat belts sometimes saved lives, sometimes killed people, and sometimes damaged people unnecessarily. A friend of mine, who had moved down south, had a son (in his twenties) who drove down to see his mother around the time of this court case and had a bad accident. He was driving on a two lane divided highway in a seriously dense fog when a semi truck hit him head on. Apparently the driver of the semi was new and in an unfamiliar area and thought that the road was a regular highway, not a divided one way. He didn't see the car until the last minute. He swerved, but still hit my friend's son's car and totaled it, destroying the passenger's side. The police told my friend's son that he was locked into place by his seat belt, which saved his life and prevented him being thrown into the damaged side of the car, but they also stated that his passenger and friend was saved because he was not wearing a seat belt. If he had been locked into the passenger's side, he would probably be dead. As you can imagine, this event caused a lot of debate about luck, fate, and guardian angels, which is why I remembered it.

I did find the following in Google, which reinforces the idea that seat belts were not the end all solution and were sometimes dangerous.

"Seat belts became mandatory with the first federal law taking effect on January 1, 1968, requiring all new cars to be equipped with seat belts. However, it wasn't until the 1980s that the use of seat belts became widely mandated and enforced, with New York being the first state to require vehicle occupants to wear them in 1984. Additionally, the National Traffic and Motor Vehicle Safety Act passed in 1966 laid the groundwork for these regulations."

"Airbags were first introduced in the US in 1970 but became a standard safety device in cars in 1998. The first car to include factory-installed airbags was the 1973 Oldsmobile Toronado. The U.S. government mandated airbags in all new passenger vehicles by 1990, emphasizing their critical role in reducing crash-related injuries and fatalities."

Less than ten years after mandating and enforcing the use of seat belts, we were looking for better ways to reduce crash-related injuries and fatalities. Airbags mitigated some of the damage caused by seat belts.

Gee

1 hour ago, Gees said:

The original case that I had mentioned was not about whether or not we should use seat belts -- it was already law that we were required to use them -- it was about advertising the seat belts. Stating that "seat belts save lives" was inaccurate, a lie, because seat belts sometimes saved lives, sometimes killed people, and sometimes damaged people unnecessarily.

But they do save lives - fewer people die when seatbelts are used. It’s a statistical argument, not a guarantee that wearing one will definitely save your life. Which is why the details matter, and would be present in a detailed summary of the case.

On 12/27/2025 at 4:16 PM, swansont said:

But they do save lives - fewer people die when seatbelts are used.

Maybe. I have driven all over the United States, in Europe, on the right side of the road and on the other side, in cars, trucks, rental vehicles, and on motorcycles. I have towed cars, trailers, boats, and campers, but I do not wear seatbelts and have not for most of my life. Yet, I am not dead. It is possible that statistically speaking, seatbelts are not good for me. 😁

On 12/27/2025 at 4:16 PM, swansont said:

It’s a statistical argument, not a guarantee that wearing one will definitely save your life.

Yes it is a statistical argument, and statistics are made up of facts -- not truths. The reality is that facts, therefore statistics, are easy to cherry pick, manipulate, and cause an invalid perception. Just look at what comes out of Washington if you want evidence of that truth. I have a lot of issues with statistics and don't trust them.

In the first place, I worked as a temp in many offices in different capacities for years, and watched clerical workers gathering information (statistics) for bosses, insurance purposes, government agencies, etc. Most workers did not give two sh*ts about the information and saw it as an added, unpaid, responsibility. Would some/many of them fudge the information if it made the job easier? Absolutely.

In this age of computers, anyone who can access information can gather statistics and thinks they are as good as an actuary -- they are not.

Then you have to look at the motivations for the gathering of the statistics. A statistical argument has the same flaws as science, in that they can both lose truth to confirmation bias.

On 12/27/2025 at 4:16 PM, swansont said:

Which is why the details matter, and would be present in a detailed summary of the case.

Yes, details matter.

In the OP, I believe that AI was supposed to be used to ensure the safety of people in vehicles. So does that mean that once the problems are identified, AI would make recommendations? Maybe suggest drivers training to individuals? Point out dangerous activities? Give advice to the drivers/passengers?

Ha Ha! I was laughing so hard, I fell off my chair. Or maybe the government agency is going to fine the drivers -- after all, they need to get their money back for purchasing AI. Yep, that is what they are going to do. It's a win win situation. The government will make money on the fines and save money because they don't have to clean up so many bodies. As for the people, yes, it may save some lives, but it will also cost them. How will that cost affect them? We don't know and I don't see anyone gathering statistics to find out. Motivation is very much a part of gathering statistics and selecting what is gathered -- which is part of why I don't trust statistics.

As to other responses, some posters thought that a person would be necessary to oversee the infractions and regulate what was fined and what was not. That would be short lived, IMO, as AI would not be cost efficient if it had to be overseen, so it would eventually be on its own, or it would be discarded.

Another poster thought that the evidence gathered by AI would be sent to the driver with the ticket, so that he could see what the problem was. This is also not likely. That evidence belongs to the prosecution/plaintiff and would not be sent to the defendant by the Court. That evidence could be solicited by the defendant though a Discovery Motion in an attempt to discover what plaintiff's evidence is, so that a defense/rebuttal could be made -- IF discovery is allowed in the case. Sometimes Discovery is not allowed between a city and a defendant. You just have to go to Court and find out what the evidence is. This is all a matter of Civil Procedures and is regulated by local laws. For example, while working in a law office in Wayne County (Detroit, Michigan) I talked to a New York attorney, who was having problems getting a child support case on the Wayne County Court docket. He had been working on it for six months and getting nowhere, so I asked him what he had Filed. After listening, I explained that we can not put a child support case on a Judge's Court docket, a Referee does that, so he needed to resend all of his information, and send second copies of everything to the Friend of the Court Referee, and be sure to include a yellow Praecipe, because without a yellow Praecipe, it would never get on the Friend of the Court docket. It would just sit in a pile of papers till the end of time. So I faxed him all the information and a month later I got an "I love New York" coffee cup in the mail.

Gee

1 hour ago, Gees said:

Maybe. I have driven all over the United States, in Europe, on the right side of the road and on the other side, in cars, trucks, rental vehicles, and on motorcycles. I have towed cars, trailers, boats, and campers, but I do not wear seatbelts and have not for most of my life. Yet, I am not dead. It is possible that statistically speaking, seatbelts are not good for me. 😁

Wow, what a crappy argument!

Just because something is not personally experienced by you in no way impacts the truth or validity of others’ experiences. And being a statistical argument, it means that you are looking at the total effect of all events.

1 hour ago, Gees said:

Yes it is a statistical argument, and statistics are made up of facts -- not truths. The reality is that facts, therefore statistics, are easy to cherry pick, manipulate, and cause an invalid perception. Just look at what comes out of Washington if you want evidence of that truth. I have a lot of issues with statistics and don't trust them.

I’m not sure that willful ignorance is something people find persuasive, especially in a science discussion setting

1 hour ago, Gees said:

In the first place, I worked as a temp in many offices in different capacities for years, and watched clerical workers gathering information (statistics) for bosses, insurance purposes, government agencies, etc. Most workers did not give two sh*ts about the information and saw it as an added, unpaid, responsibility. Would some/many of them fudge the information if it made the job easier? Absolutely.

People being lazy and/or dishonest is an indictment of those people, not statistics. If there is fault to be found you can dig into the methodology for that. But a blanket dismissal is not an argument to be taken seriously.

1 hour ago, Gees said:

Yes, details matter

And were not provided, yet the argument was made anyway.