Why are you constantly ignoring what I've said repeatedly in that is one of the reasons it isn't widely adopted as much as was anticipated by this time is exactly because they are actively addressing said issues? For example, the issue of being able to obey police hand signals?
It's one thing to feel autonomous cars aren't ready right now, but you seem to have some sort of vendetta against the idea of them ever being ready. Moving goalposts, making up stats.
AI is already better than a human in some cases, AI doesn't get sleepy, nor distracted, and has a quicker reaction time. If there are cases where, given the same conditions, a human driver would be better, then yes, the AI should be refined to meet/exceed the human driver in those situations. Having 'infallible' as the goal before mass market is unrealistic.
Automatic car designers have all the time in the world to make sure their sensors are infallible.
... and therein lies your issue. You're assuming it shouldn't exist until it's infallible. It never can be, nothing can be. They don't even need to be, they just need to be better than human drivers
The problem is that it has been demonstrated that these sensors are easily fooled. I'm sure they can see the child in broad daylight, but if there is a funny shadow that tricks the sensors then there is nothing but excuses.
Human eyes/brains can be fooled too, though that's a different discussion than something like assuming an AI can be made that would never be in any accidents ever.
They'll say "oh we didn't think we had to think of that" but they do have to think of that because they have a heavy vehicle driving itself in traffic.
Isn't that exactly why they are still testing and doing a slower than originally anticipated rollout?
Every accident where the automation kills someone will be a second class murder charge because they already had full opportunity to program that car for that situation and they hit the person anyway. As far as I know, "driving is too complicated" is not an excuse for a human getting in an accident and it shouldn't be for AI either. The AI should be expected to be able to drive safely.
There are theoretical situations (that may have actually happened?) where it would be impossible for someone to not be hurt. This is regardless of AI or experienced driver, or whatever. Something like a car is driving down the road and a person, let's say a child, suddenly comes into the road.
Options:
Stop: The child would be hit anyway due to physics, or lets say the car magically stops on a dime, it'd cause whiplash to the passenger(s)
Swerve to the right: Pedestrians on the sidewalk would be hit
Swerve to the left: Oncoming traffic would be hit.
Because we all know cars that have drivers have never once in their history ever hit someone. The result would be catastrophic and cars would be banned.
Make every single website you visit have a unique password.
Then also change those passwords often.
Do this to the point where you can't even remember the password and have to use 'reset password' anyway.
Alternatively, use a password manager and make everything depend on a single point of failure
don't permanently solder the SIM card to the device.
Removable SIM sockets are just one more thing the manufacturer has to pay for and one more reason to let the dirty customers actually remove the backplate and install one or two SIMs./s
An Embedded-SIM (eSIM) or embedded universal integrated circuit card (eUICC) is a form of programmable SIM that is embedded directly into a device.... In October 2017, Google unveiled the Pixel 2, which added eSIM support for use with its Project Fi service. The following year, Apple released the iPhone XS and iPhone XR with eSIM support.
Have a unique password for each and every site you visit, and then also change them often to the point where they aren't memorable and you end up using 'forgot password'. To get around that, use a password manager, and have everything linked to a single point of failure!
Well surely all the super-smart people here would be able to develop with their own browser that wouldn't have those issues and also be widely adopted because it'd be the best, right?
Slashdot Mirror
2nd amendment, read it.
I have, and also own 4 handguns, a shotgun, and 2 rifles. In literally the 1st 3 words of the Amendment are the words "well regulated".
That's not what "well regulated" meant in 1791.... but we'll also be hypocrites and use the modern day meaning of "arms".
So you plan to go completely off the grid and not rely on any Linux powered technology once you retire in 12 years?
I'll be dead before it gets prohibitively bad
--- Baby Boomers
Why are you constantly ignoring what I've said repeatedly in that is one of the reasons it isn't widely adopted as much as was anticipated by this time is exactly because they are actively addressing said issues? For example, the issue of being able to obey police hand signals?
It's one thing to feel autonomous cars aren't ready right now, but you seem to have some sort of vendetta against the idea of them ever being ready. Moving goalposts, making up stats.
AI is already better than a human in some cases, AI doesn't get sleepy, nor distracted, and has a quicker reaction time.
If there are cases where, given the same conditions, a human driver would be better, then yes, the AI should be refined to meet/exceed the human driver in those situations.
Having 'infallible' as the goal before mass market is unrealistic.
Because no human has ever been temporarily blinded by sun glare or oncoming traffics high beams, or heavy fog, or...
Automatic car designers have all the time in the world to make sure their sensors are infallible.
The problem is that it has been demonstrated that these sensors are easily fooled. I'm sure they can see the child in broad daylight, but if there is a funny shadow that tricks the sensors then there is nothing but excuses.
Human eyes/brains can be fooled too, though that's a different discussion than something like assuming an AI can be made that would never be in any accidents ever.
They'll say "oh we didn't think we had to think of that" but they do have to think of that because they have a heavy vehicle driving itself in traffic.
Isn't that exactly why they are still testing and doing a slower than originally anticipated rollout?
Every accident where the automation kills someone will be a second class murder charge because they already had full opportunity to program that car for that situation and they hit the person anyway. As far as I know, "driving is too complicated" is not an excuse for a human getting in an accident and it shouldn't be for AI either. The AI should be expected to be able to drive safely.
There are theoretical situations (that may have actually happened?) where it would be impossible for someone to not be hurt. This is regardless of AI or experienced driver, or whatever.
Something like a car is driving down the road and a person, let's say a child, suddenly comes into the road.
Options:
Stop: The child would be hit anyway due to physics, or lets say the car magically stops on a dime, it'd cause whiplash to the passenger(s)
Swerve to the right: Pedestrians on the sidewalk would be hit
Swerve to the left: Oncoming traffic would be hit.
Just blame the pedestrian for being in the way of the new type of transit.
Because we all know cars that have drivers have never once in their history ever hit someone. The result would be catastrophic and cars would be banned.
Those are the ones that use absolutely no JavaScript at all, just like all the best pages of the halcyon days of 90s Internet.
This is Slashdot, Microsoft can never do the right thing!
They changed the plugin interface because it was a major cause of all the memory leakage people complained about.
It's almost as if people expecting to do whatever they want to a piece of software without any repercussions is an impossible demand.
But you'd still run into the situation of figuring out just what the non-exetension version of Firefox would include.
Also, isn't that available now anyway? It's FOSS so take out as much as you want I guess?
Make every single website you visit have a unique password.
Then also change those passwords often.
Do this to the point where you can't even remember the password and have to use 'reset password' anyway.
Alternatively, use a password manager and make everything depend on a single point of failure
Until I guess the heat death of the universe.
It went all the way to 11, and no, I'm not just making a Spinal Tap joke.
don't permanently solder the SIM card to the device.
Removable SIM sockets are just one more thing the manufacturer has to pay for and one more reason to let the dirty customers actually remove the backplate and install one or two SIMs. /s
SIM that isn't physically removable you say? https://en.wikipedia.org/wiki/...
An Embedded-SIM (eSIM) or embedded universal integrated circuit card (eUICC) is a form of programmable SIM that is embedded directly into a device. ... In October 2017, Google unveiled the Pixel 2, which added eSIM support for use with its Project Fi service. The following year, Apple released the iPhone XS and iPhone XR with eSIM support.
Have a unique password for each and every site you visit, and then also change them often to the point where they aren't memorable and you end up using 'forgot password'. To get around that, use a password manager, and have everything linked to a single point of failure!
I wouldn't worry about it
We'll all be dead from catastrophic climate change by then.
Well surely all the super-smart people here would be able to develop with their own browser that wouldn't have those issues and also be widely adopted because it'd be the best, right?
According to the article, the current cost estimate is 17 G$.
17 Gillion Dollars??
M$??!!?!! MiKKKr$oft bad!!1111
MS Exec #2: Well, don't look to Slashdot for advice, because literally any decision we make will be the 'wrong' one according to them.