We are a loooooong way from a mobile/portable AI computing system that can fit in a robot.
Very true -- but we already have corporations who want to protect themselves from liability when their software does something unfortunate. With this legal innovation, they can just say "it was the robot's fault -- nothing to do with us!" and the lawsuits go away:)
Let's face it, when you consider the budgets these TLAs get to play with, they must be orders of magnitude more than theoretical physicists and mathematicians - and we already know that the NSA has more PhD mathematicians than anywhere else...
On the other hand, one thing a lot of brilliant people really want to do is become famous for answering a question or solving a problem that nobody else could -- a situation that is unlikely to happen for anyone who is working under triple-dog-secret-no-disclosure-ever-and-we-mean-it-or-else conditions at the NSA.
Any reasonably brilliant person can find a way make plenty of money, if making money is what motivates them. But the desire for money is not usually what drives brilliant people; OTOH figuring out e.g. quantum mechanics and then being permanently forbidden from ever telling anyone about their discovery would be a special kind of hell for them.
Given that, I wonder what percentage of the "best and brightest" actually end up working at secret TLAs; I suspect it might be less than you think.
Driving 1.2 miles might take 5 minutes, maximum. How much power can they possibly transfer to the vehicle battery in that time?
Also, the very first fax machine was a completely useless product, since there was nobody to send faxes to or receive faxes from. I don't know why they even bothered to manufacture it.
I like technological fixes as much as the next Slashdot reader, but I think this particular problem is a tricky one to solve with 100% reliability... and of course anything with "only" 99.9% reliability is going to get sued into the ground after the first failure, regardless of whether the failure was actually due to human error or not..
My recommendation (either instead of installing a gadget or perhaps in addition to it) would be to make sure any kid who is big enough to potentially get to the pool has also taken lessons on how to stay alive once in the pool (and ideally also how to exit the pool safely).
That way, even if the gadget fails, the kid still has a pretty good chance of survival.
The actually rules have nothing to do with distance: [link to drive handbook]
I'm not sure those are "the actual rules"; they are part of a driver handbook, which is to say, they are a common-sense guide to how to be safe driver.
All well and good, but if this matter were to go to court, I think they would be looking at what the laws say rather than one the DMV driver's handbook says, and there would be a lot less common sense involved and a lot more legalese:/
Is not what is "legal" for a blockchain what the majority of nodes maintaining the chain say is legal?
Makes perfect sense to me, in a laissez-faire kind of way.
Of course, by that same token, if the majority of the nodes cry foul the next day and accept/demand a "do over" in the form of a hard-fork, then that too is "legal" as far as the system is concerned.
If you want a really good control group, how about comparing Teslas-purchased-with-the-autopilot-option against Teslas-purchased-without-the-autopilot-option?
That's about as apples-to-apples as you're going to get, assuming there are enough of each on the road to make the comparison statistically significant (I think there probably are).
I suppose it's too obvious to point out that in a democracy, where every constituent's interests differs to some degree from every other constituent's interests, that the government's "interests" are never going to align 100% with any given person's own interests?
Whenever a choice has to be made, someone is going to be disappointed. If they are disappointed enough, they can help vote in new representatives whose policies more closely align with their own views; but of course the people who supported the original decision are also free to support its continuation. That's just democracy, not a "deep state", or a "foreign power", despite what the anti-democratic populists would have people believe.
Can we stop having versions go, 3, 3s, 4, 4s, etc. I don't get the "s", or why cannot just go 3, 4, 5, 6, etc.
The reason the model names are 'S', '3', and 'X' is so that when you've bought the whole set, it (sort of) spells out 'SEX'. Yes, really. (They originally wanted the Model 3 to be named the Model E, but Ford has the rights to the name 'Model E')
And if/when the model Y comes out, you'll be able to spell out SEXY, so there's that.
if you've seen it trying to send you into a k-rail at that bit a few times, what the fuck are you doing letting it drive on that bit, and not paying attention when it tells you to?
If he was a computer programmer, I'd say he was trying to reproduce the fault, in order to better understand the entry conditions.:/
Have you calculated how many people would die if all automobile traffic were banned, due to being unable to get access to food, medical care, their jobs, etc? Sure Theaetetus is not going to be a hypocrite, particularly when lives are at stake!
What is AI and when will we finally be only ten years away from it?
AI is the ability of a computer to perform a cognitive task that was (until recently) commonly considered to be performable only by a human mind.
Of course, the popular conception of "what is performable only by a human mind" changes from year to year, as people get used to the capabilities of technology -- that's why "playing chess at grandmaster level" and "translate text from one language to another" used to be considered "AI" but no longer are. Currently "safely drive a car on public roads" is considered AI; in ten years it will likely be considered "just another app".
So AI is here today, for certain definitions of AI. OTOH, if your definition of AI boils down to "things that computers can't do yet", then by definition that version of "AI" will always be N years away.
I call it an edge case because a person dressed in dark standing in a darkly lit section of road with an outcropping behind and choosing to walk into the path of the car is not a common event
Note that while the above is a difficult case for a human, there is really no excuse for LIDAR/Radar/Sonar sensors not to have detected this particular human. My suspicion is that we'll find out that the either some or all of those sensors weren't working properly on that car, or the car's software was inadequately tested and some sort of bug prevented it from reacting appropriately to its sensors' input. Really, "don't hit pedestrians, no matter what" should be the First Law of Automotive Robotics, and this car failed it badly.
The only problem with doing that is that you end up with a pretty mediocre electric car with lots of design compromises -- a car designed from the ground up with electric in mind will have much better range, performance, and handling. That, as much as anything, was what separated Tesla from the rest of the automobile manufacturers in terms of how its electric cars were received by the public.
If you're driving 50 MPH down a road that is signed for 50 MPH not near an intersection, and some woman runs out in front of you 20 feet away...... it's not reasonable to expect you to safely achieve the stop/avoidance that physics says your human+vehicle system is not capable of.
If you are driving close enough to other obstacles that there is a possibility of someone "appearing" just 20 feet away, then you need to drive slower. Most people subconsciously realize this and drive more slowly on narrower streets.
Safe driving is partly about planning ahead and foreseeing what could conceivably happen in the next 5-30 seconds. Not all humans are good at that, and I imagine the current round of AIs may not be particularly good at it yet either.
In fact, I see a future of bums stepping in front of the cars with the funny LIDAR units on top to get a piece of Google's net worth.
Insurance/litigation fraud is significantly more difficult to pull off when the party you're trying to extort has full video, radar, and sonar records of the incident under discussion. Your hobos are going to have to be extremely good actors.
I'm sure if it looks bad the data will be mysteriously deleted.
Only if Uber is extremely stupid and short-sighted (which is not a possibility I'm ruling out).
If Uber does the right thing and doesn't try to hide the data, then whatever problem(s) lead to the accident will get discovered, analyzed, understood, and fixed, at which point Uber can get back to testing their autonomous cars on the road again, with a (mostly) clear conscience.
If OTOH Uber surreptitiously deletes the data because it's embarrassing, then the investigators will have a very difficult time figuring out what went wrong, and therefore Uber will have a very difficult time resuming the self-driving car program. In the worst case (e.g. Uber gets caught falsifying or destroying data) they would be open to massive lawsuits and/or permanently lose their legal ability to operate in the self-driving car market.
Slashdot Mirror
We are a loooooong way from a mobile/portable AI computing system that can fit in a robot.
Very true -- but we already have corporations who want to protect themselves from liability when their software does something unfortunate. With this legal innovation, they can just say "it was the robot's fault -- nothing to do with us!" and the lawsuits go away :)
All the points you raised are addressed in the article.
Let's face it, when you consider the budgets these TLAs get to play with, they must be orders of magnitude more than theoretical physicists and mathematicians - and we already know that the NSA has more PhD mathematicians than anywhere else...
On the other hand, one thing a lot of brilliant people really want to do is become famous for answering a question or solving a problem that nobody else could -- a situation that is unlikely to happen for anyone who is working under triple-dog-secret-no-disclosure-ever-and-we-mean-it-or-else conditions at the NSA.
Any reasonably brilliant person can find a way make plenty of money, if making money is what motivates them. But the desire for money is not usually what drives brilliant people; OTOH figuring out e.g. quantum mechanics and then being permanently forbidden from ever telling anyone about their discovery would be a special kind of hell for them.
Given that, I wonder what percentage of the "best and brightest" actually end up working at secret TLAs; I suspect it might be less than you think.
We have new technology to optimize that code now. Here's the modern implementation:
int getRandom() { // generated by "spooky action at a distance"
return 4;
}
Driving 1.2 miles might take 5 minutes, maximum. How much power can they possibly transfer to the vehicle battery in that time?
Also, the very first fax machine was a completely useless product, since there was nobody to send faxes to or receive faxes from. I don't know why they even bothered to manufacture it.
They told us time and again that the cameras they put everywhere were too high up to be used for facial recognition.
The Chinese government told you that?
Did they send you a certified letter, or just drop by your house to announce it over tea?
Since when did the Chinese government care about your privacy concerns, one way or the other, even enough to lie about their intentions?
I like technological fixes as much as the next Slashdot reader, but I think this particular problem is a tricky one to solve with 100% reliability... and of course anything with "only" 99.9% reliability is going to get sued into the ground after the first failure, regardless of whether the failure was actually due to human error or not..
My recommendation (either instead of installing a gadget or perhaps in addition to it) would be to make sure any kid who is big enough to potentially get to the pool has also taken lessons on how to stay alive once in the pool (and ideally also how to exit the pool safely).
That way, even if the gadget fails, the kid still has a pretty good chance of survival.
The actually rules have nothing to do with distance: [link to drive handbook]
I'm not sure those are "the actual rules"; they are part of a driver handbook, which is to say, they are a common-sense guide to how to be safe driver.
All well and good, but if this matter were to go to court, I think they would be looking at what the laws say rather than one the DMV driver's handbook says, and there would be a lot less common sense involved and a lot more legalese :/
Is not what is "legal" for a blockchain what the majority of nodes maintaining the chain say is legal?
Makes perfect sense to me, in a laissez-faire kind of way.
Of course, by that same token, if the majority of the nodes cry foul the next day and accept/demand a "do over" in the form of a hard-fork, then that too is "legal" as far as the system is concerned.
If you want a really good control group, how about comparing Teslas-purchased-with-the-autopilot-option against Teslas-purchased-without-the-autopilot-option?
That's about as apples-to-apples as you're going to get, assuming there are enough of each on the road to make the comparison statistically significant (I think there probably are).
I suppose it's too obvious to point out that in a democracy, where every constituent's interests differs to some degree from every other constituent's interests, that the government's "interests" are never going to align 100% with any given person's own interests?
Whenever a choice has to be made, someone is going to be disappointed. If they are disappointed enough, they can help vote in new representatives whose policies more closely align with their own views; but of course the people who supported the original decision are also free to support its continuation. That's just democracy, not a "deep state", or a "foreign power", despite what the anti-democratic populists would have people believe.
Can we stop having versions go, 3, 3s, 4, 4s, etc. I don't get the "s", or why cannot just go 3, 4, 5, 6, etc.
The reason the model names are 'S', '3', and 'X' is so that when you've bought the whole set, it (sort of) spells out 'SEX'. Yes, really. (They originally wanted the Model 3 to be named the Model E, but Ford has the rights to the name 'Model E')
And if/when the model Y comes out, you'll be able to spell out SEXY, so there's that.
if you've seen it trying to send you into a k-rail at that bit a few times, what the fuck are you doing letting it drive on that bit, and not paying attention when it tells you to?
If he was a computer programmer, I'd say he was trying to reproduce the fault, in order to better understand the entry conditions. :/
Have you calculated how many people would die if all automobile traffic were banned, due to being unable to get access to food, medical care, their jobs, etc? Sure Theaetetus is not going to be a hypocrite, particularly when lives are at stake!
What is AI and when will we finally be only ten years away from it?
AI is the ability of a computer to perform a cognitive task that was (until recently) commonly considered to be performable only by a human mind.
Of course, the popular conception of "what is performable only by a human mind" changes from year to year, as people get used to the capabilities of technology -- that's why "playing chess at grandmaster level" and "translate text from one language to another" used to be considered "AI" but no longer are. Currently "safely drive a car on public roads" is considered AI; in ten years it will likely be considered "just another app".
So AI is here today, for certain definitions of AI. OTOH, if your definition of AI boils down to "things that computers can't do yet", then by definition that version of "AI" will always be N years away.
It is correct and accurate to say that most countries have *already* benefited from AI (even if it's not strong AI).
I don't doubt that's true, but did you have any specific examples in mind?
I call it an edge case because a person dressed in dark standing in a darkly lit section of road with an outcropping behind and choosing to walk into the path of the car is not a common event
Note that while the above is a difficult case for a human, there is really no excuse for LIDAR/Radar/Sonar sensors not to have detected this particular human. My suspicion is that we'll find out that the either some or all of those sensors weren't working properly on that car, or the car's software was inadequately tested and some sort of bug prevented it from reacting appropriately to its sensors' input. Really, "don't hit pedestrians, no matter what" should be the First Law of Automotive Robotics, and this car failed it badly.
It happens all the time which is why I am finally running away screaming from software.
For goodness sake, don't scream, that will only make it easier for it to hunt you down.
By noticing that people who would have bought a BMW are choosing to buy a Tesla instead. Even if they have to wait a year for their Tesla to ship.
I don't understand why existing engine bays have not been reused to fit motor/battery into existing car platforms.
This is commonly done -- in fact, it's easy enough to do that some technically-minded people like to convert their own gasoline-powered cars to electric in their garage.
The only problem with doing that is that you end up with a pretty mediocre electric car with lots of design compromises -- a car designed from the ground up with electric in mind will have much better range, performance, and handling. That, as much as anything, was what separated Tesla from the rest of the automobile manufacturers in terms of how its electric cars were received by the public.
"Operator inattention" is absolutely the ultimate goal of automated cars.
If the human has to pay attention, then the human might as well drive, and the automation is pointless.
If you're driving 50 MPH down a road that is signed for 50 MPH not near an intersection, and some woman runs out in front of you 20 feet away...... it's not reasonable to expect you to safely achieve the stop/avoidance that physics says your human+vehicle system is not capable of.
If you are driving close enough to other obstacles that there is a possibility of someone "appearing" just 20 feet away, then you need to drive slower. Most people subconsciously realize this and drive more slowly on narrower streets.
Safe driving is partly about planning ahead and foreseeing what could conceivably happen in the next 5-30 seconds. Not all humans are good at that, and I imagine the current round of AIs may not be particularly good at it yet either.
In fact, I see a future of bums stepping in front of the cars with the funny LIDAR units on top to get a piece of Google's net worth.
Insurance/litigation fraud is significantly more difficult to pull off when the party you're trying to extort has full video, radar, and sonar records of the incident under discussion. Your hobos are going to have to be extremely good actors.
I'm sure if it looks bad the data will be mysteriously deleted.
Only if Uber is extremely stupid and short-sighted (which is not a possibility I'm ruling out).
If Uber does the right thing and doesn't try to hide the data, then whatever problem(s) lead to the accident will get discovered, analyzed, understood, and fixed, at which point Uber can get back to testing their autonomous cars on the road again, with a (mostly) clear conscience.
If OTOH Uber surreptitiously deletes the data because it's embarrassing, then the investigators will have a very difficult time figuring out what went wrong, and therefore Uber will have a very difficult time resuming the self-driving car program. In the worst case (e.g. Uber gets caught falsifying or destroying data) they would be open to massive lawsuits and/or permanently lose their legal ability to operate in the self-driving car market.
Guess who will be liable.
The rental company, for renting out a car that wasn't properly maintained? Perhaps you are confusing renting with leasing.