this post was submitted on 11 Jul 2023
555 points (96.5% liked)
Technology
59689 readers
3753 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
You need to first ask yourself if it more important to put blame than to minimize risk.
"Autonomous vehicles could potentially reduce traffic fatalities by up to 90%."
"Autonomous vehicle accidents have been recorded at a slightly lower rate compared with conventional cars, at 4.7 accidents per million miles driven."
https://blog.gitnux.com/driverless-car-accident-statistics/
That opinion puts a lot of blind faith in the companies developing self driving and their infinitely altruistic motives.
That's one way of strawmanning your way out of a discussion.
It's not a strawman argument, it is a fact. Without the ability to audit the entire codebase of self-driving cars, there's no way to know if the manufacturer had knowingly hidden something in the code that might have caused accidents and fatalities too numerous to recount, but too important to ignore, that were linked to a fault in self-driving technology.
I was actually trying to find an article I'd read about Tesla's self-driving software reverting to manual control moments before impact, but I was literally flooded by fatality reports.
Strawman arguments can be factual. The entire point is that you're responding to something that wasn't the argument. You're putting words in their mouth to defeat them instead of addressing their words at face value. It is the definition of a strawman argument.
We can't audit the code for humans, but we still let them drive.
If the output for computers driving is less than for humans and the computer designers are forced to be as financially liable for car crashes as humans, why shouldn't we let computers drive?
I'm not fully in either camp in this debate, but fwiw, the humans we let drive generally suffer consequences if there is an accident due to their own negligence
Also we do audit them, it's called a license. I know it's super easy to get one in the US but in other countries they can be quite stringent.
And I'm not denying it. However, it takes a very high bar to get someone convicted of vehicular manslaughter and that usually requires evidence that the driver was grossly negligent.
If you can show that a computer can drive as well as a sober human, where is the gross negligence?
Because there's no valid excuse to prevent us from auditing their software and it could save lives. Why the hell should we allow then to use the road if they won't even let us inspect the engine?
A car isn't a human. It's a machine, and it can and should be inspected. Anything less than that is pure recklessness.
How do you think a car gets approved right now? Do we take it apart? Do we ask for the design calculations of how they designed each piece?
That isn't what happens. There is no "audit" of parts or the whole. Instead, there is a series of tests to determine road worthiness that everything in a car has to pass. We've already accepted a black box for the electronics of a car. You don't need to get approval of your code to show that pressing the brake pedal causes the brake lights turn on; they just test it to make sure that it works.
We don't audit the code already for life critical software already. It is all liability taken on by the manufacturers and verified via government testing of the finished product. What is an audit going to do when we don't it already?
It is most definitely a strawman to frame my comment as considering the companies "infinitely altruistic", no matter what lies behind the strawman. It doesn't refute my statistics but rather tries to make me look like I make an extremely silly argument I'm not making, which is the defintion of a strawman argument.
The data you cited comes straight from manufacturers, who've repeatedly been shown to lie and cherry-pick their data to intentionally mislead people about driverless car safety.
So no it's not a straw man argument at all to claim that you're putting inordinate faith in manufacturers, because that's exactly what you did. It's actually incredible to me how many of you are so irresponsible that you're not even willing to do basic cross-checking against an industry that is known for blatantly lying about safety issues.
It may be the case that every line of code of all self driving vehicles is not available for a public audit. But neither is the instruction set of every human who was taught to drive properly on the road today.
I would hope that through protesting and new legislation, that we will see the industry become more safe over time. Which we simply will never be able to achieve with human drivers.
That wasn't an opinion, it's a statistic.
No (large public) company ever has altruistic motives. They aren't inherently good or bad, just machines driven by profit.
You don't need to put faith into companies beyond the faith that is put into humans. Make companies just as financially liable as humans are, and you'll still see a decrease in accidents.
You mean those companies who will lobby and spend a fraction of their wealth to make those lawsuits disappear?
How is that different from the current system of large vehicular insurance companies spending a fraction of their wealth to make their lawsuits disappear?
It's no different at all. We should have stronger laws for such scenarios.
Ok, but in the context of letting computers drive, I feel like people want to enforce this perfect system of liability on automated systems where we already have an existing criminal and civil legal system as is that is designed to nowhere near the same standard for humans.
Why are we willing to say that it is unacceptable that no computer can kill people on the road when almost 43,000 die in the USA due to humans driving?
Uh, because software can be fixed and those deaths can be prevented? How the hell can you ask this question seriously? I can't believe how many people are willing to blatantly shill for these companies, even if it gets people fucking killed.
And no you can't claim to be saving lives because these driverless cars very often kill people in situations that a human driver would easily navigate.
And until the system is perfect, let people die on the worse system?
This isn't me shilling for a company, this is me comparing two flawed systems.
This part is bogus to me as well. My friend who used to work in self-driving said that when self driving can be "just" better than human driving, technology has won. In statistical terms, it means having slightly lesser fatalities than humans (<43k fatalities with respect to the num of human drivers).
Now it's up for debate lesser by how much exactly. Just 5% reduction or 50% reduction. If we want to go for 99% reduction, we should stop building self-driving tech altogether.
So ban all forms of driving?
You know what has much smaller fatality rates? Walking, bikes, trains and buses
So...
Your car is at fault. Their kid is dead.
Who pays for the funeral?
Does your insurance cover programming glitches?
If your insurance determined that an autonomous vehicle will cause less damage over time than a human driver, they will do that, yes.
Autonomous logic doesn't pay insurance, does it?
If so, who TF is paying the insurance behind the scenes, and who is responsible?
The owner of the vehicle is probably very openly paying.
Here's a question- if you have to agree to terms of service for the vehicle to function, and I'm guessing you would, is it really your vehicle?
We're talking about autonomous vehicles here, no driver, company owned.
So is Alphabet responsible?
Do your homework, these vehicles are owned by the parent company of Google ~~and Apple~~, Alphabet. These vehicles have no private owner. So again, who TF is responsible?
So what? It's not the gotcha you apparently believe to have found, companies can have insurance...
Companies also never seem to be held accountable. OceanGate anybody?...
That's not a good example. Courts move slow and that just barely happened and AFAIK is still being investigated (plus searching, the participants signed wavers -- though wavers don't give immunity legal negligence).
There's plenty of examples of companies being punished for negligence. It happens all the time when, say, their poorly constructed building collapses, cutting corners causes an oil spill in the Gulf of Mexico, they falsified their vehicle emissions reports, or when they abuse their market dominance.
Corporations totally do get away with a lot, but I don't see why you'd expect self driving cars to be a place where that would happen, especially since manually driven cars are already so regulated and require insurance. And everyone knows that driving is dangerous. Nobody is under any false impressions that self driving cars don't have at least some of that same danger. I mean, even if the AI was utterly perfect, they'd still need insurance to protect against cases that aren't the AI's fault.
When the vehicle disobeys orders from the police, who is at fault?
https://piped.video/watch?v=5Jev_R-JVmA
Alphabet don't own Apple.
I'll take your word on that. I've edited my comment to reflect that, but last research I did a few years ago, both companies were under the umbrella of Alphabet.
Nope, Apple have never been owned by Alphabet. Alphabet is literally just googles new parent company that they formed when restructuring.
https://en.wikipedia.org/wiki/Alphabet_Inc.
I mean, why shouldn't it? Is a programming glitch in a self driving all that different from a mechanical issue in a manually driven car?
AI driven cars are just as prone to mechanical issues as well. Is AI smart enough to deal with a flat tire? Will it pull over to the side of the road before phoning in for a mechanic, or will it just ignorantly hard stop right in the middle of the interstate?
What's AI do when there's a police officer directing traffic around an accident or through a faulty red light intersection? I've literally seen videos on that before, AI couldn't give two shits about a cop's orders as to which way to drive the vehicle.
are there actual datasets to look at and info regarding how data was collected? all the sources on that page are just domain links but don't appear to point to the data making the claims?
4.7 accidents per million miles doesn't mean much if the cars are limited to specific roads or include test tracks that give them an advantage. the degree of variance in different environments would also need to be measured such as weather effects, road conditions and traffic patterns.
I'm all for autonomous driving, but its not like companies don't fudge numbers all the time for their benefit.
Story time...
I once had a crazy accident driving only like 15-20 MPH or so down a side road, then about 20 feet in front of me some idiot backed out of his parking spot right in front of me.
Broad daylight, overcast skies, no other vehicles blocking his view even. Dude just backed up without looking like a freaking idiot.
I responded in a split second. I did not hit the brakes, as I knew I didn't have enough time or distance to stop. If I had hit the brakes, his car would have had more time to back out further and I would have smacked straight on into the passenger side of his car.
Instead of hitting the brakes, I quickly jerked the steering wheel hard and fast to the left. See, I knew an impact was inevitable at that point, I made that move to clip his bumper instead of smacking into the passenger side and ruining both vehicles.
Would an AI do that? 🤔
Would? Maybe, don't know, not sure. Could? Yes.
They tend to work on basic sensors and simplified logic. They don't tend to consider forward momentum and a vehicle pulling out perpendicular in front of you.
I believe half the programmers of autonomous vehicles never even drove a vehicle in their life.
It's weird that you think this isn't the suggested driving practice in such an instance