Move over, SawStop ...

On Thu, 23 Nov 2017 18:44:05 -0800, OFWW <[email protected]>
wrote:

>On Thu, 23 Nov 2017 11:53:47 -0800 (PST), DerbyDad03
><[email protected]> wrote:
>
>>On Thursday, November 23, 2017 at 11:40:13 AM UTC-5, OFWW wrote:
>>> On Wed, 22 Nov 2017 12:36:05 -0800 (PST), DerbyDad03
>>> <[email protected]> wrote:
>>>
>>> >On Wednesday, November 22, 2017 at 1:51:05 PM UTC-5, Ed Pawlowski wrote:
>>> >> On 11/22/2017 1:20 PM, DerbyDad03 wrote:
>>> >>
>>> >> >
>>> >> > Oh, well, no sense in waiting...
>>> >> >
>>> >> > 2nd scenario:
>>> >> >
>>> >> > 5 workers are standing on the railroad tracks. A train is heading in their
>>> >> > direction. They have no escape route. If the train continues down the tracks,
>>> >> > it will most assuredly kill them all.
>>> >> >
>>> >> > You are standing on a bridge overlooking the tracks. Next to you is a fairly
>>> >> > large person. We'll save you some trouble and let that person be a stranger.
>>> >> >
>>> >> > You have 2, and only 2, options. If you do nothing, all 5 workers will
>>> >> > be killed. If you push the stranger off the bridge, the train will kill
>>> >> > him but be stopped before the 5 workers are killed. (Don't question the
>>> >> > physics, just accept the outcome.)
>>> >> >
>>> >> > Which option do you choose?
>>> >> >
>>> >>
>>> >> I don't know. It was easy to pull the switch as there was a bit of
>>> >> disconnect there. Now it is up close and you are doing the pushing.
>>> >> One alternative is to jump yourself, but I'd not do that. Don't think I
>>> >> could push the guy either.
>>> >>
>>> >
>>> >And there in lies the rub. The "disconnected" part.
>>> >
>>> >Now, as promised, let's bring this back to technology, AI and most
>>> >certainly, people. Let's talk specifically about autonomous vehicles,
>>> >but please avoid the rabbit hole and realize that the concept applies
>>> >to just about any where AI is used and people are involved. Autonomus
>>> >vehicles (AV) are just one example.
>>> >
>>> >Imagine it's X years from now and AV's are fairly common. Imagine that an AV
>>> >is traveling down the road, with its AI in complete control of the vehicle.
>>> >The driver is using one hand get a cup of coffee from the built-in Keurig
>>> >machine and choosing a Pandora station with the other. He is completely
>>> >oblivious to what's happening outside of his vehicle.
>>> >
>>> >Now imagine that a 4 year old runs out into the road. The AI uses all of the
>>> >data at its disposal (speed, distance, weather conditions, tire pressure,
>>> >etc.) and decides that it will not be able to stop in time. It checks the
>>> >input from its 360° cameras. Can't go right because of the line of parked
>>> >cars. They won't slow the vehicle enough to avoid hitting the kid. Using
>>> >facial recognition the AI determines that the mini-van on the left contains
>>> >5 elderly people. If the AV swerves left, it will push the mini-van into
>>> >oncoming traffic, directly into the path of a 18 wheeler. The AI communicates
>>> >with the 18 wheeler's AI who responds and says "I have no place to go. If
>>> >you push the van into my lane, I'm taking out a bunch of Grandmas and
>>> >Grandpas."
>>> >
>>> >Now the AI has to make basically the same decision as in my first scenario:
>>> >Kill 1 or kill 5. For the AI, it's as easy as it was for us, right?
>>> >
>>> >"Bye Bye, kid. You should have stayed on the sidewalk."
>>> >
>>> >No emotion, right? Right, not once the AI is programmed, not once the initial
>>> >AI rules have been written, not once the facial recognition database has
>>> >been built. The question is who wrote those rules? Who decided it's OK to
>>> >kill a young kid to save the lives of 5 rickety old folks? Oh wait, maybe
>>> >it's better to save the kid and let the old folks die. They've had a full
>>> >life. Who wrote that rule? In other words, someone(s) have to decide whose
>>> >life is worth more than another's. They are essentially standing on a bridge
>>> >deciding whether to push the guy or not. They have to write the rule. They
>>> >are either going to kill the kid or push the car into the other lane.
>>> >
>>> >I, for one, don't think that I want to be sitting around that table. Having
>>> >to make the decisions would be one thing. Having to sit next to the person
>>> >that would push the guy off the bridge with a gleam in his eye would be a
>>> >totally different story.
>>>
>>> I reconsidered my thoughts on this one as well.
>>>
>>> The AV should do as it was designed to do, to the best of its
>>> capabilities. Staying in the lane when there is no option to swerve
>>> safely.
>>>
>>> There is already a legal reason for that, that being that the swerving
>>> driver assumes all the damages that incur from his action, including
>>> manslaughter.
>>
>>So in the following brake failure scenario, if the AV stays in lane and
>>kills the four "highly rated" pedestrians there are no charges, but if
>>it changes lanes and takes out the 4 slugs, jail time may ensue.
>>
>>http://static6.businessinsider.com/image/58653ba0ee14b61b008b5aea-800
>>
>>Interesting.
>
>Yes, and I've been warned that by my taking evasive action I could
>cause someone else to respond likewise and that I would he held
>accountable for what happened.

I find the assumption that a fatality involving a robot car would lead
to someone being jailed to be amusing. The people who assert this
never identify the statute under which someone would be jailed or who,
precisely this someone might be. They seem to assume that because a
human driving a car could be jailed for vehicular homicide or criminal
negligence or some such, it is automatic that someone else would be
jailed for the same offense--the trouble is that the car is legally an
inanimate object and we don't put inanimate objects in jail. So it
gets down to proving that the occupant is negligent, which is a hard
sell given that the government allowed the car to be licensed with the
understanding that it would not be controlled by the occupant, or
proving that the engineering team responsible for developing it was
negligent, which given that they can show the logic the thing used and
no doubt provide legal justification for the decision it made, will be
another tall order. So who goes to jail?

On Friday, November 24, 2017 at 10:10:01 AM UTC-5, Ed Pawlowski wrote:
> On 11/24/2017 12:37 AM, J. Clarke wrote:
>
> >>>
> >>> I find the assumption that a fatality involving a robot car would lead
> >>> to someone being jailed to be amusing. The people who assert this
> >>> never identify the statute under which someone would be jailed or who,
> >>> precisely this someone might be. They seem to assume that because a
> >>> human driving a car could be jailed for vehicular homicide or criminal
> >>> negligence or some such, it is automatic that someone else would be
> >>> jailed for the same offense--the trouble is that the car is legally an
> >>> inanimate object and we don't put inanimate objects in jail.
> >>>
> >>
>
> They can impound your car in a drug bust. Maybe they will impound your
> car for the offense. We'll build special long term impound lots for
> serious offenses, just disconnect the battery for lesser ones.
>
>
> >> You've taken it to the next level, into the real word scenario and out
> >> of the programming stage.
> >>
> >> Personally I would assume that anything designed would have to
> >> co-exist with real world laws and responsibilities. Even the final
> >> owner could be held responsible. See the laws regarding experimental
> >> aircraft, hang gliders, etc.
> >
> > Experimental aircraft and hang gliders are controlled by a human. If
> > they are involved in a fatl accident, the operator gets scrutinized.
> > An autonomous car is not under human control, it is its own operator,
> > the occupant is a passenger.
>
> The programmer will be jailed. Or maybe they will stick a pin in a
> Voodoo doll to punish him.
>
>
> >
> > We don't have "real world law" governing fatalities involving
> > autonomous vehicles. The engineering would, initially (I hope) be
> > based on existing case law involving human drivers and what the courts
> > held that they should or should not have done in particular
> > situations. But there won't be any actual law until either the
> > legislatures write statutes or the courts issue rulings, and the
> > latter won't happen until there are such vehicles in service in
> > sufficient quantity to generate cases.
>
> The sensible thing would be to gather the most brilliant minds of the TV
> ambulance chasing lawyers and let them come up with guidelines for
> liability. Can you think of anything more fair than that?

Sure. Build a random number generator into the AI. The AI simply uses the
random number to decide who to take out at the time of the incident.

"Step right up, spin the wheel, take your chances."

It'll all be "hit or miss" so to speak.

On 11/24/2017 12:37 AM, J. Clarke wrote:

>>>
>>> I find the assumption that a fatality involving a robot car would lead
>>> to someone being jailed to be amusing. The people who assert this
>>> never identify the statute under which someone would be jailed or who,
>>> precisely this someone might be. They seem to assume that because a
>>> human driving a car could be jailed for vehicular homicide or criminal
>>> negligence or some such, it is automatic that someone else would be
>>> jailed for the same offense--the trouble is that the car is legally an
>>> inanimate object and we don't put inanimate objects in jail.
>>>
>>

They can impound your car in a drug bust. Maybe they will impound your
car for the offense. We'll build special long term impound lots for
serious offenses, just disconnect the battery for lesser ones.


>> You've taken it to the next level, into the real word scenario and out
>> of the programming stage.
>>
>> Personally I would assume that anything designed would have to
>> co-exist with real world laws and responsibilities. Even the final
>> owner could be held responsible. See the laws regarding experimental
>> aircraft, hang gliders, etc.
>
> Experimental aircraft and hang gliders are controlled by a human. If
> they are involved in a fatl accident, the operator gets scrutinized.
> An autonomous car is not under human control, it is its own operator,
> the occupant is a passenger.

The programmer will be jailed. Or maybe they will stick a pin in a
Voodoo doll to punish him.


>
> We don't have "real world law" governing fatalities involving
> autonomous vehicles. The engineering would, initially (I hope) be
> based on existing case law involving human drivers and what the courts
> held that they should or should not have done in particular
> situations. But there won't be any actual law until either the
> legislatures write statutes or the courts issue rulings, and the
> latter won't happen until there are such vehicles in service in
> sufficient quantity to generate cases.

The sensible thing would be to gather the most brilliant minds of the TV
ambulance chasing lawyers and let them come up with guidelines for
liability. Can you think of anything more fair than that?

On Thu, 23 Nov 2017 20:52:09 -0800, OFWW <[email protected]>
wrote:

>On Thu, 23 Nov 2017 23:10:05 -0500, J. Clarke
><[email protected]> wrote:
>
>>On Thu, 23 Nov 2017 18:44:05 -0800, OFWW <[email protected]>
>>wrote:
>>
>>>On Thu, 23 Nov 2017 11:53:47 -0800 (PST), DerbyDad03
>>><[email protected]> wrote:
>>>
>>>>On Thursday, November 23, 2017 at 11:40:13 AM UTC-5, OFWW wrote:
>>>>> On Wed, 22 Nov 2017 12:36:05 -0800 (PST), DerbyDad03
>>>>> <[email protected]> wrote:
>>>>>
>>>>> >On Wednesday, November 22, 2017 at 1:51:05 PM UTC-5, Ed Pawlowski wrote:
>>>>> >> On 11/22/2017 1:20 PM, DerbyDad03 wrote:
>>>>> >>
>>>>> >> >
>>>>> >> > Oh, well, no sense in waiting...
>>>>> >> >
>>>>> >> > 2nd scenario:
>>>>> >> >
>>>>> >> > 5 workers are standing on the railroad tracks. A train is heading in their
>>>>> >> > direction. They have no escape route. If the train continues down the tracks,
>>>>> >> > it will most assuredly kill them all.
>>>>> >> >
>>>>> >> > You are standing on a bridge overlooking the tracks. Next to you is a fairly
>>>>> >> > large person. We'll save you some trouble and let that person be a stranger.
>>>>> >> >
>>>>> >> > You have 2, and only 2, options. If you do nothing, all 5 workers will
>>>>> >> > be killed. If you push the stranger off the bridge, the train will kill
>>>>> >> > him but be stopped before the 5 workers are killed. (Don't question the
>>>>> >> > physics, just accept the outcome.)
>>>>> >> >
>>>>> >> > Which option do you choose?
>>>>> >> >
>>>>> >>
>>>>> >> I don't know. It was easy to pull the switch as there was a bit of
>>>>> >> disconnect there. Now it is up close and you are doing the pushing.
>>>>> >> One alternative is to jump yourself, but I'd not do that. Don't think I
>>>>> >> could push the guy either.
>>>>> >>
>>>>> >
>>>>> >And there in lies the rub. The "disconnected" part.
>>>>> >
>>>>> >Now, as promised, let's bring this back to technology, AI and most
>>>>> >certainly, people. Let's talk specifically about autonomous vehicles,
>>>>> >but please avoid the rabbit hole and realize that the concept applies
>>>>> >to just about any where AI is used and people are involved. Autonomus
>>>>> >vehicles (AV) are just one example.
>>>>> >
>>>>> >Imagine it's X years from now and AV's are fairly common. Imagine that an AV
>>>>> >is traveling down the road, with its AI in complete control of the vehicle.
>>>>> >The driver is using one hand get a cup of coffee from the built-in Keurig
>>>>> >machine and choosing a Pandora station with the other. He is completely
>>>>> >oblivious to what's happening outside of his vehicle.
>>>>> >
>>>>> >Now imagine that a 4 year old runs out into the road. The AI uses all of the
>>>>> >data at its disposal (speed, distance, weather conditions, tire pressure,
>>>>> >etc.) and decides that it will not be able to stop in time. It checks the
>>>>> >input from its 360° cameras. Can't go right because of the line of parked
>>>>> >cars. They won't slow the vehicle enough to avoid hitting the kid. Using
>>>>> >facial recognition the AI determines that the mini-van on the left contains
>>>>> >5 elderly people. If the AV swerves left, it will push the mini-van into
>>>>> >oncoming traffic, directly into the path of a 18 wheeler. The AI communicates
>>>>> >with the 18 wheeler's AI who responds and says "I have no place to go. If
>>>>> >you push the van into my lane, I'm taking out a bunch of Grandmas and
>>>>> >Grandpas."
>>>>> >
>>>>> >Now the AI has to make basically the same decision as in my first scenario:
>>>>> >Kill 1 or kill 5. For the AI, it's as easy as it was for us, right?
>>>>> >
>>>>> >"Bye Bye, kid. You should have stayed on the sidewalk."
>>>>> >
>>>>> >No emotion, right? Right, not once the AI is programmed, not once the initial
>>>>> >AI rules have been written, not once the facial recognition database has
>>>>> >been built. The question is who wrote those rules? Who decided it's OK to
>>>>> >kill a young kid to save the lives of 5 rickety old folks? Oh wait, maybe
>>>>> >it's better to save the kid and let the old folks die. They've had a full
>>>>> >life. Who wrote that rule? In other words, someone(s) have to decide whose
>>>>> >life is worth more than another's. They are essentially standing on a bridge
>>>>> >deciding whether to push the guy or not. They have to write the rule. They
>>>>> >are either going to kill the kid or push the car into the other lane.
>>>>> >
>>>>> >I, for one, don't think that I want to be sitting around that table. Having
>>>>> >to make the decisions would be one thing. Having to sit next to the person
>>>>> >that would push the guy off the bridge with a gleam in his eye would be a
>>>>> >totally different story.
>>>>>
>>>>> I reconsidered my thoughts on this one as well.
>>>>>
>>>>> The AV should do as it was designed to do, to the best of its
>>>>> capabilities. Staying in the lane when there is no option to swerve
>>>>> safely.
>>>>>
>>>>> There is already a legal reason for that, that being that the swerving
>>>>> driver assumes all the damages that incur from his action, including
>>>>> manslaughter.
>>>>
>>>>So in the following brake failure scenario, if the AV stays in lane and
>>>>kills the four "highly rated" pedestrians there are no charges, but if
>>>>it changes lanes and takes out the 4 slugs, jail time may ensue.
>>>>
>>>>http://static6.businessinsider.com/image/58653ba0ee14b61b008b5aea-800
>>>>
>>>>Interesting.
>>>
>>>Yes, and I've been warned that by my taking evasive action I could
>>>cause someone else to respond likewise and that I would he held
>>>accountable for what happened.
>>
>>I find the assumption that a fatality involving a robot car would lead
>>to someone being jailed to be amusing. The people who assert this
>>never identify the statute under which someone would be jailed or who,
>>precisely this someone might be. They seem to assume that because a
>>human driving a car could be jailed for vehicular homicide or criminal
>>negligence or some such, it is automatic that someone else would be
>>jailed for the same offense--the trouble is that the car is legally an
>>inanimate object and we don't put inanimate objects in jail. So it
>>gets down to proving that the occupant is negligent, which is a hard
>>sell given that the government allowed the car to be licensed with the
>>understanding that it would not be controlled by the occupant, or
>>proving that the engineering team responsible for developing it was
>>negligent, which given that they can show the logic the thing used and
>>no doubt provide legal justification for the decision it made, will be
>>another tall order. So who goes to jail?
>>
>
>You've taken it to the next level, into the real word scenario and out
>of the programming stage.
>
>Personally I would assume that anything designed would have to
>co-exist with real world laws and responsibilities. Even the final
>owner could be held responsible. See the laws regarding experimental
>aircraft, hang gliders, etc.

Experimental aircraft and hang gliders are controlled by a human. If
they are involved in a fatl accident, the operator gets scrutinized.
An autonomous car is not under human control, it is its own operator,
the occupant is a passenger.

We don't have "real world law" governing fatalities involving
autonomous vehicles. The engineering would, initially (I hope) be
based on existing case law involving human drivers and what the courts
held that they should or should not have done in particular
situations. But there won't be any actual law until either the
legislatures write statutes or the courts issue rulings, and the
latter won't happen until there are such vehicles in service in
sufficient quantity to generate cases.

Rather than hang gliders and homebuilts, consider a Globalhawk that
hits an airliner. Assuming no negligence on the part of the airliner
crew, who do you go after? Do you go after the Air Force, Northrop
Grumman, Raytheon, or somebody else? And of what are they likely to
be found guilty?

>But we should be sticking to this hypothetical example given us.

It was suggested that someone would go to jail. I still want to know
who and what crime they committed.

On Thu, 23 Nov 2017 23:10:05 -0500, J. Clarke
<[email protected]> wrote:

>On Thu, 23 Nov 2017 18:44:05 -0800, OFWW <[email protected]>
>wrote:
>
>>On Thu, 23 Nov 2017 11:53:47 -0800 (PST), DerbyDad03
>><[email protected]> wrote:
>>
>>>On Thursday, November 23, 2017 at 11:40:13 AM UTC-5, OFWW wrote:
>>>> On Wed, 22 Nov 2017 12:36:05 -0800 (PST), DerbyDad03
>>>> <[email protected]> wrote:
>>>>
>>>> >On Wednesday, November 22, 2017 at 1:51:05 PM UTC-5, Ed Pawlowski wrote:
>>>> >> On 11/22/2017 1:20 PM, DerbyDad03 wrote:
>>>> >>
>>>> >> >
>>>> >> > Oh, well, no sense in waiting...
>>>> >> >
>>>> >> > 2nd scenario:
>>>> >> >
>>>> >> > 5 workers are standing on the railroad tracks. A train is heading in their
>>>> >> > direction. They have no escape route. If the train continues down the tracks,
>>>> >> > it will most assuredly kill them all.
>>>> >> >
>>>> >> > You are standing on a bridge overlooking the tracks. Next to you is a fairly
>>>> >> > large person. We'll save you some trouble and let that person be a stranger.
>>>> >> >
>>>> >> > You have 2, and only 2, options. If you do nothing, all 5 workers will
>>>> >> > be killed. If you push the stranger off the bridge, the train will kill
>>>> >> > him but be stopped before the 5 workers are killed. (Don't question the
>>>> >> > physics, just accept the outcome.)
>>>> >> >
>>>> >> > Which option do you choose?
>>>> >> >
>>>> >>
>>>> >> I don't know. It was easy to pull the switch as there was a bit of
>>>> >> disconnect there. Now it is up close and you are doing the pushing.
>>>> >> One alternative is to jump yourself, but I'd not do that. Don't think I
>>>> >> could push the guy either.
>>>> >>
>>>> >
>>>> >And there in lies the rub. The "disconnected" part.
>>>> >
>>>> >Now, as promised, let's bring this back to technology, AI and most
>>>> >certainly, people. Let's talk specifically about autonomous vehicles,
>>>> >but please avoid the rabbit hole and realize that the concept applies
>>>> >to just about any where AI is used and people are involved. Autonomus
>>>> >vehicles (AV) are just one example.
>>>> >
>>>> >Imagine it's X years from now and AV's are fairly common. Imagine that an AV
>>>> >is traveling down the road, with its AI in complete control of the vehicle.
>>>> >The driver is using one hand get a cup of coffee from the built-in Keurig
>>>> >machine and choosing a Pandora station with the other. He is completely
>>>> >oblivious to what's happening outside of his vehicle.
>>>> >
>>>> >Now imagine that a 4 year old runs out into the road. The AI uses all of the
>>>> >data at its disposal (speed, distance, weather conditions, tire pressure,
>>>> >etc.) and decides that it will not be able to stop in time. It checks the
>>>> >input from its 360° cameras. Can't go right because of the line of parked
>>>> >cars. They won't slow the vehicle enough to avoid hitting the kid. Using
>>>> >facial recognition the AI determines that the mini-van on the left contains
>>>> >5 elderly people. If the AV swerves left, it will push the mini-van into
>>>> >oncoming traffic, directly into the path of a 18 wheeler. The AI communicates
>>>> >with the 18 wheeler's AI who responds and says "I have no place to go. If
>>>> >you push the van into my lane, I'm taking out a bunch of Grandmas and
>>>> >Grandpas."
>>>> >
>>>> >Now the AI has to make basically the same decision as in my first scenario:
>>>> >Kill 1 or kill 5. For the AI, it's as easy as it was for us, right?
>>>> >
>>>> >"Bye Bye, kid. You should have stayed on the sidewalk."
>>>> >
>>>> >No emotion, right? Right, not once the AI is programmed, not once the initial
>>>> >AI rules have been written, not once the facial recognition database has
>>>> >been built. The question is who wrote those rules? Who decided it's OK to
>>>> >kill a young kid to save the lives of 5 rickety old folks? Oh wait, maybe
>>>> >it's better to save the kid and let the old folks die. They've had a full
>>>> >life. Who wrote that rule? In other words, someone(s) have to decide whose
>>>> >life is worth more than another's. They are essentially standing on a bridge
>>>> >deciding whether to push the guy or not. They have to write the rule. They
>>>> >are either going to kill the kid or push the car into the other lane.
>>>> >
>>>> >I, for one, don't think that I want to be sitting around that table. Having
>>>> >to make the decisions would be one thing. Having to sit next to the person
>>>> >that would push the guy off the bridge with a gleam in his eye would be a
>>>> >totally different story.
>>>>
>>>> I reconsidered my thoughts on this one as well.
>>>>
>>>> The AV should do as it was designed to do, to the best of its
>>>> capabilities. Staying in the lane when there is no option to swerve
>>>> safely.
>>>>
>>>> There is already a legal reason for that, that being that the swerving
>>>> driver assumes all the damages that incur from his action, including
>>>> manslaughter.
>>>
>>>So in the following brake failure scenario, if the AV stays in lane and
>>>kills the four "highly rated" pedestrians there are no charges, but if
>>>it changes lanes and takes out the 4 slugs, jail time may ensue.
>>>
>>>http://static6.businessinsider.com/image/58653ba0ee14b61b008b5aea-800
>>>
>>>Interesting.
>>
>>Yes, and I've been warned that by my taking evasive action I could
>>cause someone else to respond likewise and that I would he held
>>accountable for what happened.
>
>I find the assumption that a fatality involving a robot car would lead
>to someone being jailed to be amusing. The people who assert this
>never identify the statute under which someone would be jailed or who,
>precisely this someone might be. They seem to assume that because a
>human driving a car could be jailed for vehicular homicide or criminal
>negligence or some such, it is automatic that someone else would be
>jailed for the same offense--the trouble is that the car is legally an
>inanimate object and we don't put inanimate objects in jail. So it
>gets down to proving that the occupant is negligent, which is a hard
>sell given that the government allowed the car to be licensed with the
>understanding that it would not be controlled by the occupant, or
>proving that the engineering team responsible for developing it was
>negligent, which given that they can show the logic the thing used and
>no doubt provide legal justification for the decision it made, will be
>another tall order. So who goes to jail?
>

You've taken it to the next level, into the real word scenario and out
of the programming stage.

Personally I would assume that anything designed would have to
co-exist with real world laws and responsibilities. Even the final
owner could be held responsible. See the laws regarding experimental
aircraft, hang gliders, etc.

But we should be sticking to this hypothetical example given us.

On Thu, 23 Nov 2017 23:10:05 -0500, J. Clarke
<[email protected]> wrote:

>On Thu, 23 Nov 2017 18:44:05 -0800, OFWW <[email protected]>
>wrote:
>
>>On Thu, 23 Nov 2017 11:53:47 -0800 (PST), DerbyDad03
>><[email protected]> wrote:
>>
>>>On Thursday, November 23, 2017 at 11:40:13 AM UTC-5, OFWW wrote:
>>>> On Wed, 22 Nov 2017 12:36:05 -0800 (PST), DerbyDad03
>>>> <[email protected]> wrote:
>>>>
>>>> >On Wednesday, November 22, 2017 at 1:51:05 PM UTC-5, Ed Pawlowski wrote:
>>>> >> On 11/22/2017 1:20 PM, DerbyDad03 wrote:
>>>> >>
>>>> >> >
>>>> >> > Oh, well, no sense in waiting...
>>>> >> >
>>>> >> > 2nd scenario:
>>>> >> >
>>>> >> > 5 workers are standing on the railroad tracks. A train is heading in their
>>>> >> > direction. They have no escape route. If the train continues down the tracks,
>>>> >> > it will most assuredly kill them all.
>>>> >> >
>>>> >> > You are standing on a bridge overlooking the tracks. Next to you is a fairly
>>>> >> > large person. We'll save you some trouble and let that person be a stranger.
>>>> >> >
>>>> >> > You have 2, and only 2, options. If you do nothing, all 5 workers will
>>>> >> > be killed. If you push the stranger off the bridge, the train will kill
>>>> >> > him but be stopped before the 5 workers are killed. (Don't question the
>>>> >> > physics, just accept the outcome.)
>>>> >> >
>>>> >> > Which option do you choose?
>>>> >> >
>>>> >>
>>>> >> I don't know. It was easy to pull the switch as there was a bit of
>>>> >> disconnect there. Now it is up close and you are doing the pushing.
>>>> >> One alternative is to jump yourself, but I'd not do that. Don't think I
>>>> >> could push the guy either.
>>>> >>
>>>> >
>>>> >And there in lies the rub. The "disconnected" part.
>>>> >
>>>> >Now, as promised, let's bring this back to technology, AI and most
>>>> >certainly, people. Let's talk specifically about autonomous vehicles,
>>>> >but please avoid the rabbit hole and realize that the concept applies
>>>> >to just about any where AI is used and people are involved. Autonomus
>>>> >vehicles (AV) are just one example.
>>>> >
>>>> >Imagine it's X years from now and AV's are fairly common. Imagine that an AV
>>>> >is traveling down the road, with its AI in complete control of the vehicle.
>>>> >The driver is using one hand get a cup of coffee from the built-in Keurig
>>>> >machine and choosing a Pandora station with the other. He is completely
>>>> >oblivious to what's happening outside of his vehicle.
>>>> >
>>>> >Now imagine that a 4 year old runs out into the road. The AI uses all of the
>>>> >data at its disposal (speed, distance, weather conditions, tire pressure,
>>>> >etc.) and decides that it will not be able to stop in time. It checks the
>>>> >input from its 360° cameras. Can't go right because of the line of parked
>>>> >cars. They won't slow the vehicle enough to avoid hitting the kid. Using
>>>> >facial recognition the AI determines that the mini-van on the left contains
>>>> >5 elderly people. If the AV swerves left, it will push the mini-van into
>>>> >oncoming traffic, directly into the path of a 18 wheeler. The AI communicates
>>>> >with the 18 wheeler's AI who responds and says "I have no place to go. If
>>>> >you push the van into my lane, I'm taking out a bunch of Grandmas and
>>>> >Grandpas."
>>>> >
>>>> >Now the AI has to make basically the same decision as in my first scenario:
>>>> >Kill 1 or kill 5. For the AI, it's as easy as it was for us, right?
>>>> >
>>>> >"Bye Bye, kid. You should have stayed on the sidewalk."
>>>> >
>>>> >No emotion, right? Right, not once the AI is programmed, not once the initial
>>>> >AI rules have been written, not once the facial recognition database has
>>>> >been built. The question is who wrote those rules? Who decided it's OK to
>>>> >kill a young kid to save the lives of 5 rickety old folks? Oh wait, maybe
>>>> >it's better to save the kid and let the old folks die. They've had a full
>>>> >life. Who wrote that rule? In other words, someone(s) have to decide whose
>>>> >life is worth more than another's. They are essentially standing on a bridge
>>>> >deciding whether to push the guy or not. They have to write the rule. They
>>>> >are either going to kill the kid or push the car into the other lane.
>>>> >
>>>> >I, for one, don't think that I want to be sitting around that table. Having
>>>> >to make the decisions would be one thing. Having to sit next to the person
>>>> >that would push the guy off the bridge with a gleam in his eye would be a
>>>> >totally different story.
>>>>
>>>> I reconsidered my thoughts on this one as well.
>>>>
>>>> The AV should do as it was designed to do, to the best of its
>>>> capabilities. Staying in the lane when there is no option to swerve
>>>> safely.
>>>>
>>>> There is already a legal reason for that, that being that the swerving
>>>> driver assumes all the damages that incur from his action, including
>>>> manslaughter.
>>>
>>>So in the following brake failure scenario, if the AV stays in lane and
>>>kills the four "highly rated" pedestrians there are no charges, but if
>>>it changes lanes and takes out the 4 slugs, jail time may ensue.
>>>
>>>http://static6.businessinsider.com/image/58653ba0ee14b61b008b5aea-800
>>>
>>>Interesting.
>>
>>Yes, and I've been warned that by my taking evasive action I could
>>cause someone else to respond likewise and that I would he held
>>accountable for what happened.
>
>I find the assumption that a fatality involving a robot car would lead
>to someone being jailed to be amusing. The people who assert this
>never identify the statute under which someone would be jailed or who,
>precisely this someone might be. They seem to assume that because a
>human driving a car could be jailed for vehicular homicide or criminal
>negligence or some such, it is automatic that someone else would be
>jailed for the same offense--the trouble is that the car is legally an
>inanimate object and we don't put inanimate objects in jail. So it
>gets down to proving that the occupant is negligent, which is a hard
>sell given that the government allowed the car to be licensed with the
>understanding that it would not be controlled by the occupant, or
>proving that the engineering team responsible for developing it was
>negligent, which given that they can show the logic the thing used and
>no doubt provide legal justification for the decision it made, will be
>another tall order. So who goes to jail?
>
The software developer who signed off on the failing module.

On Fri, 24 Nov 2017 11:58:06 -0600, Markem <[email protected]>
wrote:

>On Fri, 24 Nov 2017 00:53:07 -0500, J. Clarke
><[email protected]> wrote:
>
>>On Thu, 23 Nov 2017 23:46:52 -0600, Markem <[email protected]>
>>wrote:
>>
>>>On Fri, 24 Nov 2017 00:37:20 -0500, J. Clarke
>>><[email protected]> wrote:
>>>
>>>>It was suggested that someone would go to jail. I still want to know
>>>>who and what crime they committed.
>>>
>>>Damages would be a tort case,
>>
>>So why do you mention damages?
>>
>>> as to who and what crime that would be
>>>determined in court. Some DA looking for publicty would brings
>>>charges.
>>
>>What charges? To bring charges there must have been a chargeable
>>offense, which means that a plausible argument can be made that some
>>law was violated. So what law do you believe would have been
>>violated? Or do you just _like_ being laughed out of court?
>
>I am not looking for political office, ever heard the saying a DA can
>indict a ham sandwich.

But when was the last time a ham sandwich was imprisoned?

On Fri, 24 Nov 2017 18:09:51 -0800, OFWW <[email protected]>
wrote:

>On Fri, 24 Nov 2017 09:11:16 -0500, J. Clarke
><[email protected]> wrote:
>
>>>>>But we should be sticking to this hypothetical example given us.
>>>>
>>>>It was suggested that someone would go to jail. I still want to know
>>>>who and what crime they committed.
>>>
>>>The person who did not stay in their own lane, and ended up committing
>>>involuntary manslaughter.
>>
>>Are you arguing that an autonomous vehicle is a "person"? You
>>really don't seem to grasp the concept. Rather than a car with an
>>occupant, make it a car, say a robot taxicab, that is going somewhere
>>or other unoccupied.
>>
>
>Is not a "who" a person? and yes, I realize the optimum goal is for a
>stand alone vehicle independent of owner operator. The robotic taxicab
>is already in test mode.
>
>>>In the case you bring up the AV can be currently over ridden at
>>>anytime by the occupant. There are already AV vehicles operating on
>>>the streets.
>>
>>In what case that I bring up?
>
>The case of the option for switching lanes. Your questioning as who
>can be at fault. I brought up the fact that experiment air craft have
>a lifetime indebtedness going back to the original maker and designer.
>It was to answer just who was culpable.

Check your attributions. There are many people participating in this
discussion. I did not bring up that case.

>> Globalhawk doesn't _have_ an occupant.
>>(when people use words with which you are unfamiliar, you should at
>>least Google those words before opining). There are very few
>>autonomous vehicles and currently they are for the most part operated
>>with a safety driver, but that is not anybody's long-term plan. Google
>>already has at least one demonstrator with no steering wheel or pedals
>>and Uber is planning on using driverless cars in their ride sharing
>>service--ultimately those would also have no controls accessible to
>>the passenger.
>>
>
>There are a lot

For certain rather small values of "lot".

> of autonomous vehicles running around, it just depends
>are where you are, some have already been in real world accidents,

Yes, mostly other vehicles hitting them. I believe that there has
been one Google car collision that was attributed to decisionmaking by
the software. I'm ignoring the Tesla incident because that is not
supposed to be a completely autonomous system.

>Uber already were testing vehicles but required a person in the case
>just in case.

I believe it is the government requring the person.

>And yes, I knew globalhawks do not have an occupant resident in the
>vehicle, but they are all monitored.

What do you mean when you say "monitored"? A human has to detect that
there is a danger, turn off the robot, and take control. If the robot
does not know that there is a danger it is unlikely that the human
will have any more information than the robot does.

>As to vehicles some have a safety
>driver and some do not. The globalhawks have built in sensory devices
>themselves for alarming, etc. and all the data from radar, satellites
>etc. The info for the full technology that they and the operators have
>is not disclosed. Plus it is a secret as to who all are operating the
>vehicles so the bottom line would be the government operating them.

So you're saying that the entire government would go to jail? Dream
on.

>But thank you for your comment on my knowledge and how to fix it. :)
>
>>>Regarding your "whose at fault" scenario, just look at the court cases
>>>against gun makers, as if guns kill people.
>>
>>I have not introduced a "who's at fault scenariao". I have asked what
>>law would be violated and who would be jailed. "At fault" decides who
>>pays damages, not who goes to jail. I am not discussing damages, I am
>>discussing JAIL. You do know what a jail is, do you not?
>>
>
>Sorry, my Internet connection is down and I cannot google it.

And yet you can post here.

>>>So can we know return to the question or at the least, wood working?
>>
>>You're the one who started feeding the troll.
>
>Sorry, I am not privy to the list, so I'll just make this my last post
>on the subject, but I will read your reply.

Hope springs eternal.

On Wed, 22 Nov 2017 21:06:38 +0000, Spalted Walt
<[email protected]> wrote:

>DerbyDad03 <[email protected]> wrote:
>
>> On Wednesday, November 22, 2017 at 1:51:05 PM UTC-5, Ed Pawlowski wrote:
>> > On 11/22/2017 1:20 PM, DerbyDad03 wrote:
>> >
>> > >
>> > > Oh, well, no sense in waiting...
>> > >
>> > > 2nd scenario:
>> > >
>> > > 5 workers are standing on the railroad tracks. A train is heading in their
>> > > direction. They have no escape route. If the train continues down the tracks,
>> > > it will most assuredly kill them all.
>> > >
>> > > You are standing on a bridge overlooking the tracks. Next to you is a fairly
>> > > large person. We'll save you some trouble and let that person be a stranger.
>> > >
>> > > You have 2, and only 2, options. If you do nothing, all 5 workers will
>> > > be killed. If you push the stranger off the bridge, the train will kill
>> > > him but be stopped before the 5 workers are killed. (Don't question the
>> > > physics, just accept the outcome.)
>> > >
>> > > Which option do you choose?
>> > >
>> >
>> > I don't know. It was easy to pull the switch as there was a bit of
>> > disconnect there. Now it is up close and you are doing the pushing.
>> > One alternative is to jump yourself, but I'd not do that. Don't think I
>> > could push the guy either.
>> >
>>
>> And there in lies the rub. The "disconnected" part.
>>
>> Now, as promised, let's bring this back to technology, AI and most
>> certainly, people. Let's talk specifically about autonomous vehicles,
>> but please avoid the rabbit hole and realize that the concept applies
>> to just about any where AI is used and people are involved. Autonomus
>> vehicles (AV) are just one example.
>>
>> Imagine it's X years from now and AV's are fairly common. Imagine that an AV
>> is traveling down the road, with its AI in complete control of the vehicle.
>> The driver is using one hand get a cup of coffee from the built-in Keurig
>> machine and choosing a Pandora station with the other. He is completely
>> oblivious to what's happening outside of his vehicle.
>>
>> Now imagine that a 4 year old runs out into the road. The AI uses all of the
>> data at its disposal (speed, distance, weather conditions, tire pressure,
>> etc.) and decides that it will not be able to stop in time. It checks the
>> input from its 360° cameras. Can't go right because of the line of parked
>> cars. They won't slow the vehicle enough to avoid hitting the kid. Using
>> facial recognition the AI determines that the mini-van on the left contains
>> 5 elderly people. If the AV swerves left, it will push the mini-van into
>> oncoming traffic, directly into the path of a 18 wheeler. The AI communicates
>> with the 18 wheeler's AI who responds and says "I have no place to go. If
>> you push the van into my lane, I'm taking out a bunch of Grandmas and
>> Grandpas."

The problem with this scenario is that it assumes that the AI has only
human eyes for sensors. It sees the four year old on radar near the
side of the road, detects a possible hazard, and slows down before
arriving near the four year old.
>>
>> Now the AI has to make basically the same decision as in my first scenario:
>> Kill 1 or kill 5. For the AI, it's as easy as it was for us, right?
>>
>> "Bye Bye, kid. You should have stayed on the sidewalk."
>>
>> No emotion, right? Right, not once the AI is programmed, not once the initial
>> AI rules have been written, not once the facial recognition database has
>> been built. The question is who wrote those rules? Who decided it's OK to
>> kill a young kid to save the lives of 5 rickety old folks? Oh wait, maybe
>> it's better to save the kid and let the old folks die. They've had a full
>> life. Who wrote that rule? In other words, someone(s) have to decide whose
>> life is worth more than another's. They are essentially standing on a bridge
>> deciding whether to push the guy or not. They have to write the rule. They
>> are either going to kill the kid or push the car into the other lane.
>>
>> I, for one, don't think that I want to be sitting around that table. Having
>> to make the decisions would be one thing. Having to sit next to the person
>> that would push the guy off the bridge with a gleam in his eye would be a
>> totally different story.
>
>https://pbs.twimg.com/media/Cp0D5oCWIAAxSUT.jpg
>
>LOL!

On Wednesday, November 22, 2017 at 6:38:28 PM UTC-5, J. Clarke wrote:

...snip...

> The problem with this scenario is that it assumes that the AI has only
> human eyes for sensors. It sees the four year old on radar near the
> side of the road, detects a possible hazard, and slows down before
> arriving near the four year old.
> >>=20

OK, have it your way.

"To truly guarantee a pedestrian=E2=80=99s safety, an AV would have to slow=
to a=20
crawl any time a pedestrian is walking nearby on a sidewalk, in case the=20
pedestrian decided to throw themselves in front of the vehicle," Noah=20
Goodall, a scientist with the Virginia Transportation Research Council,=20
wrote by email."

http://www.businessinsider.com/self-driving-cars-already-deciding-who-to-ki=
ll-2016-12

...snip...

On Wed, 22 Nov 2017 21:06:38 +0000, Spalted Walt
<[email protected]> wrote:

>DerbyDad03 <[email protected]> wrote:
>
>> On Wednesday, November 22, 2017 at 1:51:05 PM UTC-5, Ed Pawlowski wrote:
>> > On 11/22/2017 1:20 PM, DerbyDad03 wrote:
>> >
>> > >
>> > > Oh, well, no sense in waiting...
>> > >
>> > > 2nd scenario:
>> > >
>> > > 5 workers are standing on the railroad tracks. A train is heading in their
>> > > direction. They have no escape route. If the train continues down the tracks,
>> > > it will most assuredly kill them all.
>> > >
>> > > You are standing on a bridge overlooking the tracks. Next to you is a fairly
>> > > large person. We'll save you some trouble and let that person be a stranger.
>> > >
>> > > You have 2, and only 2, options. If you do nothing, all 5 workers will
>> > > be killed. If you push the stranger off the bridge, the train will kill
>> > > him but be stopped before the 5 workers are killed. (Don't question the
>> > > physics, just accept the outcome.)
>> > >
>> > > Which option do you choose?
>> > >
>> >
>> > I don't know. It was easy to pull the switch as there was a bit of
>> > disconnect there. Now it is up close and you are doing the pushing.
>> > One alternative is to jump yourself, but I'd not do that. Don't think I
>> > could push the guy either.
>> >
>>
>> And there in lies the rub. The "disconnected" part.
>>
>> Now, as promised, let's bring this back to technology, AI and most
>> certainly, people. Let's talk specifically about autonomous vehicles,
>> but please avoid the rabbit hole and realize that the concept applies
>> to just about any where AI is used and people are involved. Autonomus
>> vehicles (AV) are just one example.
>>
>> Imagine it's X years from now and AV's are fairly common. Imagine that an AV
>> is traveling down the road, with its AI in complete control of the vehicle.
>> The driver is using one hand get a cup of coffee from the built-in Keurig
>> machine and choosing a Pandora station with the other. He is completely
>> oblivious to what's happening outside of his vehicle.
>>
>> Now imagine that a 4 year old runs out into the road. The AI uses all of the
>> data at its disposal (speed, distance, weather conditions, tire pressure,
>> etc.) and decides that it will not be able to stop in time. It checks the
>> input from its 360° cameras. Can't go right because of the line of parked
>> cars. They won't slow the vehicle enough to avoid hitting the kid. Using
>> facial recognition the AI determines that the mini-van on the left contains
>> 5 elderly people. If the AV swerves left, it will push the mini-van into
>> oncoming traffic, directly into the path of a 18 wheeler. The AI communicates
>> with the 18 wheeler's AI who responds and says "I have no place to go. If
>> you push the van into my lane, I'm taking out a bunch of Grandmas and
>> Grandpas."
>>
>> Now the AI has to make basically the same decision as in my first scenario:
>> Kill 1 or kill 5. For the AI, it's as easy as it was for us, right?
>>
>> "Bye Bye, kid. You should have stayed on the sidewalk."
>>
>> No emotion, right? Right, not once the AI is programmed, not once the initial
>> AI rules have been written, not once the facial recognition database has
>> been built. The question is who wrote those rules? Who decided it's OK to
>> kill a young kid to save the lives of 5 rickety old folks? Oh wait, maybe
>> it's better to save the kid and let the old folks die. They've had a full
>> life. Who wrote that rule? In other words, someone(s) have to decide whose
>> life is worth more than another's. They are essentially standing on a bridge
>> deciding whether to push the guy or not. They have to write the rule. They
>> are either going to kill the kid or push the car into the other lane.
>>
>> I, for one, don't think that I want to be sitting around that table. Having
>> to make the decisions would be one thing. Having to sit next to the person
>> that would push the guy off the bridge with a gleam in his eye would be a
>> totally different story.
>
>https://pbs.twimg.com/media/Cp0D5oCWIAAxSUT.jpg

ROTF

>LOL!

On Wednesday, November 22, 2017 at 6:38:28 PM UTC-5, J. Clarke wrote:
> On Wed, 22 Nov 2017 21:06:38 +0000, Spalted Walt
> <[email protected]> wrote:
>=20
> >DerbyDad03 <[email protected]> wrote:
> >
> >> On Wednesday, November 22, 2017 at 1:51:05 PM UTC-5, Ed Pawlowski wrot=
e:
> >> > On 11/22/2017 1:20 PM, DerbyDad03 wrote:
> >> >=20
> >> > >=20
> >> > > Oh, well, no sense in waiting...
> >> > >=20
> >> > > 2nd scenario:
> >> > >=20
> >> > > 5 workers are standing on the railroad tracks. A train is heading =
in their
> >> > > direction. They have no escape route. If the train continues down =
the tracks,
> >> > > it will most assuredly kill them all.
> >> > >=20
> >> > > You are standing on a bridge overlooking the tracks. Next to you i=
s a fairly
> >> > > large person. We'll save you some trouble and let that person be a=
stranger.
> >> > >=20
> >> > > You have 2, and only 2, options. If you do nothing, all 5 workers =
will
> >> > > be killed. If you push the stranger off the bridge, the train will=
kill
> >> > > him but be stopped before the 5 workers are killed. (Don't questio=
n the
> >> > > physics, just accept the outcome.)
> >> > >=20
> >> > > Which option do you choose?
> >> > >=20
> >> >=20
> >> > I don't know. It was easy to pull the switch as there was a bit of=
=20
> >> > disconnect there. Now it is up close and you are doing the pushing.=
=20
> >> > One alternative is to jump yourself, but I'd not do that. Don't thi=
nk I=20
> >> > could push the guy either.
> >> >=20
> >>=20
> >> And there in lies the rub. The "disconnected" part.
> >>=20
> >> Now, as promised, let's bring this back to technology, AI and most=20
> >> certainly, people. Let's talk specifically about autonomous vehicles,=
=20
> >> but please avoid the rabbit hole and realize that the concept applies
> >> to just about any where AI is used and people are involved. Autonomus
> >> vehicles (AV) are just one example.
> >>=20
> >> Imagine it's X years from now and AV's are fairly common. Imagine that=
an AV=20
> >> is traveling down the road, with its AI in complete control of the veh=
icle.=20
> >> The driver is using one hand get a cup of coffee from the built-in Keu=
rig=20
> >> machine and choosing a Pandora station with the other. He is completel=
y=20
> >> oblivious to what's happening outside of his vehicle.=20
> >>=20
> >> Now imagine that a 4 year old runs out into the road. The AI uses all =
of the=20
> >> data at its disposal (speed, distance, weather conditions, tire pressu=
re,=20
> >> etc.) and decides that it will not be able to stop in time. It checks =
the=20
> >> input from its 360=C2=B0 cameras. Can't go right because of the line o=
f parked=20
> >> cars. They won't slow the vehicle enough to avoid hitting the kid. Usi=
ng=20
> >> facial recognition the AI determines that the mini-van on the left con=
tains=20
> >> 5 elderly people. If the AV swerves left, it will push the mini-van in=
to=20
> >> oncoming traffic, directly into the path of a 18 wheeler. The AI commu=
nicates=20
> >> with the 18 wheeler's AI who responds and says "I have no place to go.=
If=20
> >> you push the van into my lane, I'm taking out a bunch of Grandmas and=
=20
> >> Grandpas."
>=20
> The problem with this scenario is that it assumes that the AI has only
> human eyes for sensors. It sees the four year old on radar near the
> side of the road, detects a possible hazard, and slows down before
> arriving near the four year old.

Gee, I don't know which of my 2 comments to post first...

No, the problem is that you did not read the description of the scenario
carefully enough. "Can't go right because of the line of parked=20
cars." Unless the radar is airborne or can see through metal, it won't
detect the kid in time.

But - and this is big but - it ain't about the seeing of the kid or not.=20
You could shoot a thousand arrows through the the scenario. It was never
meant to be perfect. The point is that there are still humans involved=20
that have to write the rules related which person or persons to kill.

People have to decide when it's OK to push the big guy off the bridge.

On Nov 24, 2017, J. Clarke wrote
(in article<[email protected]>):

> On Fri, 24 Nov 2017 18:39:03 -0500, Joseph Gwinn
> <[email protected]> wrote:
>
> > On Nov 24, 2017, J. Clarke wrote
> > (in article<[email protected]>):
> >
> > > On Fri, 24 Nov 2017 11:33:41 -0500, Joseph Gwinn
> > > <[email protected]> wrote:
> > >
> > > > On Nov 24, 2017, OFWW wrote
> > > > (in article<[email protected]>):
> > > >
> > > > > On Fri, 24 Nov 2017 00:37:20 -0500, J. Clarke
> > > > > <[email protected]> wrote:
> > > > >
> > > > > > On Thu, 23 Nov 2017 20:52:09 -0800, OFWW<[email protected]>
> > > > > > wrote:
> > > > > >
> > > > > > > On Thu, 23 Nov 2017 23:10:05 -0500, J. Clarke
> > > > > > > <[email protected]> wrote:
> > > > > > >
> > > > > > > > On Thu, 23 Nov 2017 18:44:05 -0800, OFWW<[email protected]>
> > > > > > > > wrote:
> > > > > > > >
> > > > > > > > > On Thu, 23 Nov 2017 11:53:47 -0800 (PST), DerbyDad03
> > > > > > > > > <[email protected]> wrote:
> > > > > > > > >
> > > > > > > > > > On Thursday, November 23, 2017 at 11:40:13 AM UTC-5, OFWW wrote:
> > > > > > > > > > > On Wed, 22 Nov 2017 12:36:05 -0800 (PST), DerbyDad03
> > > > > > > > > > > <[email protected]> wrote:
> > > > > > > > > > >
> > > > > > > > > > > > On Wednesday, November 22, 2017 at 1:51:05 PM UTC-5, Ed Pawlowski
> > > > > > > > > > > > wrote:
> > > > > > > > > > > > > On 11/22/2017 1:20 PM, DerbyDad03 wrote:
> > > > > > > > > > > > >
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > Oh, well, no sense in waiting...
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > 2nd scenario:
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > 5 workers are standing on the railroad tracks. A train is
> > > > > > > > > > > > > > heading
> > > > > > > > > > > > > > in their
> > > > > > > > > > > > > > direction. They have no escape route. If the train continues
> > > > > > > > > > > > > > down
> > > > > > > > > > > > > > the tracks,
> > > > > > > > > > > > > > it will most assuredly kill them all.
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > You are standing on a bridge overlooking the tracks. Next to you
> > > > > > > > > > > > > > is
> > > > > > > > > > > > > > a fairly
> > > > > > > > > > > > > > large person. We'll save you some trouble and let that person
> > > > > > > > > > > > > > be a
> > > > > > > > > > > > > > stranger.
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > You have 2, and only 2, options. If you do nothing, all 5
> > > > > > > > > > > > > > workers
> > > > > > > > > > > > > > will
> > > > > > > > > > > > > > be killed. If you push the stranger off the bridge, the train
> > > > > > > > > > > > > > will
> > > > > > > > > > > > > > kill
> > > > > > > > > > > > > > him but be stopped before the 5 workers are killed. (Don't
> > > > > > > > > > > > > > question
> > > > > > > > > > > > > > the
> > > > > > > > > > > > > > physics, just accept the outcome.)
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > Which option do you choose?
> > > > > > > > > > > > >
> > > > > > > > > > > > > I don't know. It was easy to pull the switch as there was a bit
> > > > > > > > > > > > > of
> > > > > > > > > > > > > disconnect there. Now it is up close and you are doing the
> > > > > > > > > > > > > pushing.
> > > > > > > > > > > > > One alternative is to jump yourself, but I'd not do that. Don't
> > > > > > > > > > > > > think I
> > > > > > > > > > > > > could push the guy either.
> > > > > > > > > > > >
> > > > > > > > > > > > And there in lies the rub. The "disconnected" part.
> > > > > > > > > > > >
> > > > > > > > > > > > Now, as promised, let's bring this back to technology, AI and most
> > > > > > > > > > > > certainly, people. Let's talk specifically about autonomous
> > > > > > > > > > > > vehicles,
> > > > > > > > > > > > but please avoid the rabbit hole and realize that the concept
> > > > > > > > > > > > applies
> > > > > > > > > > > > to just about any where AI is used and people are involved.
> > > > > > > > > > > > Autonomus
> > > > > > > > > > > > vehicles (AV) are just one example.
> > > > > > > > > > > >
> > > > > > > > > > > > Imagine it's X years from now and AV's are fairly common. Imagine
> > > > > > > > > > > > that an AV
> > > > > > > > > > > > is traveling down the road, with its AI in complete control of the
> > > > > > > > > > > > vehicle.
> > > > > > > > > > > > The driver is using one hand get a cup of coffee from the built-in
> > > > > > > > > > > > Keurig
> > > > > > > > > > > > machine and choosing a Pandora station with the other. He is
> > > > > > > > > > > > completely
> > > > > > > > > > > > oblivious to what's happening outside of his vehicle.
> > > > > > > > > > > >
> > > > > > > > > > > > Now imagine that a 4 year old runs out into the road. The AI uses
> > > > > > > > > > > > all
> > > > > > > > > > > > of the
> > > > > > > > > > > > data at its disposal (speed, distance, weather conditions, tire
> > > > > > > > > > > > pressure,
> > > > > > > > > > > > etc.) and decides that it will not be able to stop in time. It
> > > > > > > > > > > > checks
> > > > > > > > > > > > the
> > > > > > > > > > > > input from its 360° cameras. Can't go right because of the line
> > > > > > > > > > > > of
> > > > > > > > > > > > parked
> > > > > > > > > > > > cars. They won't slow the vehicle enough to avoid hitting the kid.
> > > > > > > > > > > > Using
> > > > > > > > > > > > facial recognition the AI determines that the mini-van on the left
> > > > > > > > > > > > contains
> > > > > > > > > > > > 5 elderly people. If the AV swerves left, it will push the
> > > > > > > > > > > > mini-van
> > > > > > > > > > > > into
> > > > > > > > > > > > oncoming traffic, directly into the path of a 18 wheeler. The AI
> > > > > > > > > > > > communicates
> > > > > > > > > > > > with the 18 wheeler's AI who responds and says "I have no place to
> > > > > > > > > > > > go. If
> > > > > > > > > > > > you push the van into my lane, I'm taking out a bunch of Grandmas
> > > > > > > > > > > > and
> > > > > > > > > > > > Grandpas."
> > > > > > > > > > > >
> > > > > > > > > > > > Now the AI has to make basically the same decision as in my first
> > > > > > > > > > > > scenario:
> > > > > > > > > > > > Kill 1 or kill 5. For the AI, it's as easy as it was for us,
> > > > > > > > > > > > right?
> > > > > > > > > > > >
> > > > > > > > > > > > "Bye Bye, kid. You should have stayed on the sidewalk."
> > > > > > > > > > > >
> > > > > > > > > > > > No emotion, right? Right, not once the AI is programmed, not once
> > > > > > > > > > > > the
> > > > > > > > > > > > initial
> > > > > > > > > > > > AI rules have been written, not once the facial recognition
> > > > > > > > > > > > database
> > > > > > > > > > > > has
> > > > > > > > > > > > been built. The question is who wrote those rules? Who decided
> > > > > > > > > > > > it's
> > > > > > > > > > > > OK to
> > > > > > > > > > > > kill a young kid to save the lives of 5 rickety old folks? Oh
> > > > > > > > > > > > wait,
> > > > > > > > > > > > maybe
> > > > > > > > > > > > it's better to save the kid and let the old folks die. They've
> > > > > > > > > > > > had a
> > > > > > > > > > > > full
> > > > > > > > > > > > life. Who wrote that rule? In other words, someone(s) have to
> > > > > > > > > > > > decide
> > > > > > > > > > > > whose
> > > > > > > > > > > > life is worth more than another's. They are essentially standing
> > > > > > > > > > > > on
> > > > > > > > > > > > a
> > > > > > > > > > > > bridge
> > > > > > > > > > > > deciding whether to push the guy or not. They have to write the
> > > > > > > > > > > > rule.
> > > > > > > > > > > > They
> > > > > > > > > > > > are either going to kill the kid or push the car into the other
> > > > > > > > > > > > lane.
> > > > > > > > > > > >
> > > > > > > > > > > > I, for one, don't think that I want to be sitting around that
> > > > > > > > > > > > table.
> > > > > > > > > > > > Having
> > > > > > > > > > > > to make the decisions would be one thing. Having to sit next to
> > > > > > > > > > > > the
> > > > > > > > > > > > person
> > > > > > > > > > > > that would push the guy off the bridge with a gleam in his eye
> > > > > > > > > > > > would
> > > > > > > > > > > > be a
> > > > > > > > > > > > totally different story.
> > > > > > > > > > >
> > > > > > > > > > > I reconsidered my thoughts on this one as well.
> > > > > > > > > > >
> > > > > > > > > > > The AV should do as it was designed to do, to the best of its
> > > > > > > > > > > capabilities. Staying in the lane when there is no option to swerve
> > > > > > > > > > > safely.
> > > > > > > > > > >
> > > > > > > > > > > There is already a legal reason for that, that being that the
> > > > > > > > > > > swerving
> > > > > > > > > > > driver assumes all the damages that incur from his action,
> > > > > > > > > > > including
> > > > > > > > > > > manslaughter.
> > > > > > > > > >
> > > > > > > > > > So in the following brake failure scenario, if the AV stays in lane
> > > > > > > > > > and
> > > > > > > > > > kills the four "highly rated" pedestrians there are no charges, but
> > > > > > > > > > if
> > > > > > > > > > it changes lanes and takes out the 4 slugs, jail time may ensue.
> > > > > > > > > >
> > > > > > > > > > http://static6.businessinsider.com/image/58653ba0ee14b61b008b5aea-80
> > > > > > > > > > 0
> > > > > > > > > >
> > > > > > > > > > Interesting.
> > > > > > > > >
> > > > > > > > > Yes, and I've been warned that by my taking evasive action I could
> > > > > > > > > cause someone else to respond likewise and that I would he held
> > > > > > > > > accountable for what happened.
> > > > > > > >
> > > > > > > > I find the assumption that a fatality involving a robot car would lead
> > > > > > > > to someone being jailed to be amusing. The people who assert this
> > > > > > > > never identify the statute under which someone would be jailed or who,
> > > > > > > > precisely this someone might be. They seem to assume that because a
> > > > > > > > human driving a car could be jailed for vehicular homicide or criminal
> > > > > > > > negligence or some such, it is automatic that someone else would be
> > > > > > > > jailed for the same offense--the trouble is that the car is legally an
> > > > > > > > inanimate object and we don't put inanimate objects in jail. So it
> > > > > > > > gets down to proving that the occupant is negligent, which is a hard
> > > > > > > > sell given that the government allowed the car to be licensed with the
> > > > > > > > understanding that it would not be controlled by the occupant, or
> > > > > > > > proving that the engineering team responsible for developing it was
> > > > > > > > negligent, which given that they can show the logic the thing used and
> > > > > > > > no doubt provide legal justification for the decision it made, will be
> > > > > > > > another tall order. So who goes to jail?
> > > > > > >
> > > > > > > You've taken it to the next level, into the real word scenario and out
> > > > > > > of the programming stage.
> > > > > > >
> > > > > > > Personally I would assume that anything designed would have to
> > > > > > > co-exist with real world laws and responsibilities. Even the final
> > > > > > > owner could be held responsible. See the laws regarding experimental
> > > > > > > aircraft, hang gliders, etc.
> > > > > >
> > > > > > Experimental aircraft and hang gliders are controlled by a human. If
> > > > > > they are involved in a fatl accident, the operator gets scrutinized.
> > > > > > An autonomous car is not under human control, it is its own operator,
> > > > > > the occupant is a passenger.
> > > > > >
> > > > > > We don't have "real world law" governing fatalities involving
> > > > > > autonomous vehicles. The engineering would, initially (I hope) be
> > > > > > based on existing case law involving human drivers and what the courts
> > > > > > held that they should or should not have done in particular
> > > > > > situations. But there won't be any actual law until either the
> > > > > > legislatures write statutes or the courts issue rulings, and the
> > > > > > latter won't happen until there are such vehicles in service in
> > > > > > sufficient quantity to generate cases.
> > > > > >
> > > > > > Rather than hang gliders and homebuilts, consider a Globalhawk that
> > > > > > hits an airliner. Assuming no negligence on the part of the airliner
> > > > > > crew, who do you go after? Do you go after the Air Force, Northrop
> > > > > > Grumman, Raytheon, or somebody else? And of what are they likely to
> > > > > > be found guilty?
> > > >
> > > > GlobalHawk drones do have human pilots. Although they are not on board,
> > > > they
> > > > are in control via a stellite link and can be thousands of miles away.
> > > >
> > > > .<http://www.aviationtoday.com/2017/03/16/day-life-us-air-force-drone-pilo
> > > > t/
> > >
> > > You are conflating Reaper and Globalhawk and totally missing the
> > > point.
> >
> > Could you be more specific? Exactly what is wrong?
>
> Reaper is a combat drone and is normally operated manually. We don't
> let robots decided to shoot people yet. Globalhawk is a recon drone
> and is normally autonomous. It has no weapons so shooting people is
> not an issue. It can be operated manually and normally is in high
> traffic areas for exactly the "what if it hits an airliner" reason,
> but for most of its mission profile it is autonomous.

So GlobalHawk is autonomous in the same sense as an airliner under autopilot
during the long flight to and from the theater. It is the human pilot who is
responsible for the whole flight.

.
> The article mentions Globalhawk in passing but then goes on to spend
> the rest of its time discussing piloting Predator, which while still
> in the inventory is ancestral to Reaper.

Yep.

Joe Gwinn

On Fri, 24 Nov 2017 16:23:59 -0500, J. Clarke
<[email protected]> wrote:

>On Fri, 24 Nov 2017 11:58:06 -0600, Markem <[email protected]>
>wrote:
>
>>On Fri, 24 Nov 2017 00:53:07 -0500, J. Clarke
>><[email protected]> wrote:
>>
>>>On Thu, 23 Nov 2017 23:46:52 -0600, Markem <[email protected]>
>>>wrote:
>>>
>>>>On Fri, 24 Nov 2017 00:37:20 -0500, J. Clarke
>>>><[email protected]> wrote:
>>>>
>>>>>It was suggested that someone would go to jail. I still want to know
>>>>>who and what crime they committed.
>>>>
>>>>Damages would be a tort case,
>>>
>>>So why do you mention damages?
>>>
>>>> as to who and what crime that would be
>>>>determined in court. Some DA looking for publicty would brings
>>>>charges.
>>>
>>>What charges? To bring charges there must have been a chargeable
>>>offense, which means that a plausible argument can be made that some
>>>law was violated. So what law do you believe would have been
>>>violated? Or do you just _like_ being laughed out of court?
>>
>>I am not looking for political office, ever heard the saying a DA can
>>indict a ham sandwich.
>
>But when was the last time a ham sandwich was imprisoned?

It transformed into a penicillin based mold and could no longer be
held.

On Fri, 24 Nov 2017 18:39:03 -0500, Joseph Gwinn
<[email protected]> wrote:

>On Nov 24, 2017, J. Clarke wrote
>(in article<[email protected]>):
>
>> On Fri, 24 Nov 2017 11:33:41 -0500, Joseph Gwinn
>> <[email protected]> wrote:
>>
>> > On Nov 24, 2017, OFWW wrote
>> > (in article<[email protected]>):
>> >
>> > > On Fri, 24 Nov 2017 00:37:20 -0500, J. Clarke
>> > > <[email protected]> wrote:
>> > >
>> > > > On Thu, 23 Nov 2017 20:52:09 -0800, OFWW<[email protected]>
>> > > > wrote:
>> > > >
>> > > > > On Thu, 23 Nov 2017 23:10:05 -0500, J. Clarke
>> > > > > <[email protected]> wrote:
>> > > > >
>> > > > > > On Thu, 23 Nov 2017 18:44:05 -0800, OFWW<[email protected]>
>> > > > > > wrote:
>> > > > > >
>> > > > > > > On Thu, 23 Nov 2017 11:53:47 -0800 (PST), DerbyDad03
>> > > > > > > <[email protected]> wrote:
>> > > > > > >
>> > > > > > > > On Thursday, November 23, 2017 at 11:40:13 AM UTC-5, OFWW wrote:
>> > > > > > > > > On Wed, 22 Nov 2017 12:36:05 -0800 (PST), DerbyDad03
>> > > > > > > > > <[email protected]> wrote:
>> > > > > > > > >
>> > > > > > > > > > On Wednesday, November 22, 2017 at 1:51:05 PM UTC-5, Ed Pawlowski
>> > > > > > > > > > wrote:
>> > > > > > > > > > > On 11/22/2017 1:20 PM, DerbyDad03 wrote:
>> > > > > > > > > > >
>> > > > > > > > > > > >
>> > > > > > > > > > > > Oh, well, no sense in waiting...
>> > > > > > > > > > > >
>> > > > > > > > > > > > 2nd scenario:
>> > > > > > > > > > > >
>> > > > > > > > > > > > 5 workers are standing on the railroad tracks. A train is heading
>> > > > > > > > > > > > in their
>> > > > > > > > > > > > direction. They have no escape route. If the train continues down
>> > > > > > > > > > > > the tracks,
>> > > > > > > > > > > > it will most assuredly kill them all.
>> > > > > > > > > > > >
>> > > > > > > > > > > > You are standing on a bridge overlooking the tracks. Next to you
>> > > > > > > > > > > > is
>> > > > > > > > > > > > a fairly
>> > > > > > > > > > > > large person. We'll save you some trouble and let that person be a
>> > > > > > > > > > > > stranger.
>> > > > > > > > > > > >
>> > > > > > > > > > > > You have 2, and only 2, options. If you do nothing, all 5 workers
>> > > > > > > > > > > > will
>> > > > > > > > > > > > be killed. If you push the stranger off the bridge, the train will
>> > > > > > > > > > > > kill
>> > > > > > > > > > > > him but be stopped before the 5 workers are killed. (Don't
>> > > > > > > > > > > > question
>> > > > > > > > > > > > the
>> > > > > > > > > > > > physics, just accept the outcome.)
>> > > > > > > > > > > >
>> > > > > > > > > > > > Which option do you choose?
>> > > > > > > > > > >
>> > > > > > > > > > > I don't know. It was easy to pull the switch as there was a bit of
>> > > > > > > > > > > disconnect there. Now it is up close and you are doing the pushing.
>> > > > > > > > > > > One alternative is to jump yourself, but I'd not do that. Don't
>> > > > > > > > > > > think I
>> > > > > > > > > > > could push the guy either.
>> > > > > > > > > >
>> > > > > > > > > > And there in lies the rub. The "disconnected" part.
>> > > > > > > > > >
>> > > > > > > > > > Now, as promised, let's bring this back to technology, AI and most
>> > > > > > > > > > certainly, people. Let's talk specifically about autonomous
>> > > > > > > > > > vehicles,
>> > > > > > > > > > but please avoid the rabbit hole and realize that the concept
>> > > > > > > > > > applies
>> > > > > > > > > > to just about any where AI is used and people are involved.
>> > > > > > > > > > Autonomus
>> > > > > > > > > > vehicles (AV) are just one example.
>> > > > > > > > > >
>> > > > > > > > > > Imagine it's X years from now and AV's are fairly common. Imagine
>> > > > > > > > > > that an AV
>> > > > > > > > > > is traveling down the road, with its AI in complete control of the
>> > > > > > > > > > vehicle.
>> > > > > > > > > > The driver is using one hand get a cup of coffee from the built-in
>> > > > > > > > > > Keurig
>> > > > > > > > > > machine and choosing a Pandora station with the other. He is
>> > > > > > > > > > completely
>> > > > > > > > > > oblivious to what's happening outside of his vehicle.
>> > > > > > > > > >
>> > > > > > > > > > Now imagine that a 4 year old runs out into the road. The AI uses
>> > > > > > > > > > all
>> > > > > > > > > > of the
>> > > > > > > > > > data at its disposal (speed, distance, weather conditions, tire
>> > > > > > > > > > pressure,
>> > > > > > > > > > etc.) and decides that it will not be able to stop in time. It
>> > > > > > > > > > checks
>> > > > > > > > > > the
>> > > > > > > > > > input from its 360° cameras. Can't go right because of the line of
>> > > > > > > > > > parked
>> > > > > > > > > > cars. They won't slow the vehicle enough to avoid hitting the kid.
>> > > > > > > > > > Using
>> > > > > > > > > > facial recognition the AI determines that the mini-van on the left
>> > > > > > > > > > contains
>> > > > > > > > > > 5 elderly people. If the AV swerves left, it will push the mini-van
>> > > > > > > > > > into
>> > > > > > > > > > oncoming traffic, directly into the path of a 18 wheeler. The AI
>> > > > > > > > > > communicates
>> > > > > > > > > > with the 18 wheeler's AI who responds and says "I have no place to
>> > > > > > > > > > go. If
>> > > > > > > > > > you push the van into my lane, I'm taking out a bunch of Grandmas
>> > > > > > > > > > and
>> > > > > > > > > > Grandpas."
>> > > > > > > > > >
>> > > > > > > > > > Now the AI has to make basically the same decision as in my first
>> > > > > > > > > > scenario:
>> > > > > > > > > > Kill 1 or kill 5. For the AI, it's as easy as it was for us, right?
>> > > > > > > > > >
>> > > > > > > > > > "Bye Bye, kid. You should have stayed on the sidewalk."
>> > > > > > > > > >
>> > > > > > > > > > No emotion, right? Right, not once the AI is programmed, not once
>> > > > > > > > > > the
>> > > > > > > > > > initial
>> > > > > > > > > > AI rules have been written, not once the facial recognition database
>> > > > > > > > > > has
>> > > > > > > > > > been built. The question is who wrote those rules? Who decided it's
>> > > > > > > > > > OK to
>> > > > > > > > > > kill a young kid to save the lives of 5 rickety old folks? Oh wait,
>> > > > > > > > > > maybe
>> > > > > > > > > > it's better to save the kid and let the old folks die. They've had a
>> > > > > > > > > > full
>> > > > > > > > > > life. Who wrote that rule? In other words, someone(s) have to decide
>> > > > > > > > > > whose
>> > > > > > > > > > life is worth more than another's. They are essentially standing on
>> > > > > > > > > > a
>> > > > > > > > > > bridge
>> > > > > > > > > > deciding whether to push the guy or not. They have to write the
>> > > > > > > > > > rule.
>> > > > > > > > > > They
>> > > > > > > > > > are either going to kill the kid or push the car into the other
>> > > > > > > > > > lane.
>> > > > > > > > > >
>> > > > > > > > > > I, for one, don't think that I want to be sitting around that table.
>> > > > > > > > > > Having
>> > > > > > > > > > to make the decisions would be one thing. Having to sit next to the
>> > > > > > > > > > person
>> > > > > > > > > > that would push the guy off the bridge with a gleam in his eye would
>> > > > > > > > > > be a
>> > > > > > > > > > totally different story.
>> > > > > > > > >
>> > > > > > > > > I reconsidered my thoughts on this one as well.
>> > > > > > > > >
>> > > > > > > > > The AV should do as it was designed to do, to the best of its
>> > > > > > > > > capabilities. Staying in the lane when there is no option to swerve
>> > > > > > > > > safely.
>> > > > > > > > >
>> > > > > > > > > There is already a legal reason for that, that being that the
>> > > > > > > > > swerving
>> > > > > > > > > driver assumes all the damages that incur from his action, including
>> > > > > > > > > manslaughter.
>> > > > > > > >
>> > > > > > > > So in the following brake failure scenario, if the AV stays in lane
>> > > > > > > > and
>> > > > > > > > kills the four "highly rated" pedestrians there are no charges, but if
>> > > > > > > > it changes lanes and takes out the 4 slugs, jail time may ensue.
>> > > > > > > >
>> > > > > > > > http://static6.businessinsider.com/image/58653ba0ee14b61b008b5aea-800
>> > > > > > > >
>> > > > > > > > Interesting.
>> > > > > > >
>> > > > > > > Yes, and I've been warned that by my taking evasive action I could
>> > > > > > > cause someone else to respond likewise and that I would he held
>> > > > > > > accountable for what happened.
>> > > > > >
>> > > > > > I find the assumption that a fatality involving a robot car would lead
>> > > > > > to someone being jailed to be amusing. The people who assert this
>> > > > > > never identify the statute under which someone would be jailed or who,
>> > > > > > precisely this someone might be. They seem to assume that because a
>> > > > > > human driving a car could be jailed for vehicular homicide or criminal
>> > > > > > negligence or some such, it is automatic that someone else would be
>> > > > > > jailed for the same offense--the trouble is that the car is legally an
>> > > > > > inanimate object and we don't put inanimate objects in jail. So it
>> > > > > > gets down to proving that the occupant is negligent, which is a hard
>> > > > > > sell given that the government allowed the car to be licensed with the
>> > > > > > understanding that it would not be controlled by the occupant, or
>> > > > > > proving that the engineering team responsible for developing it was
>> > > > > > negligent, which given that they can show the logic the thing used and
>> > > > > > no doubt provide legal justification for the decision it made, will be
>> > > > > > another tall order. So who goes to jail?
>> > > > >
>> > > > > You've taken it to the next level, into the real word scenario and out
>> > > > > of the programming stage.
>> > > > >
>> > > > > Personally I would assume that anything designed would have to
>> > > > > co-exist with real world laws and responsibilities. Even the final
>> > > > > owner could be held responsible. See the laws regarding experimental
>> > > > > aircraft, hang gliders, etc.
>> > > >
>> > > > Experimental aircraft and hang gliders are controlled by a human. If
>> > > > they are involved in a fatl accident, the operator gets scrutinized.
>> > > > An autonomous car is not under human control, it is its own operator,
>> > > > the occupant is a passenger.
>> > > >
>> > > > We don't have "real world law" governing fatalities involving
>> > > > autonomous vehicles. The engineering would, initially (I hope) be
>> > > > based on existing case law involving human drivers and what the courts
>> > > > held that they should or should not have done in particular
>> > > > situations. But there won't be any actual law until either the
>> > > > legislatures write statutes or the courts issue rulings, and the
>> > > > latter won't happen until there are such vehicles in service in
>> > > > sufficient quantity to generate cases.
>> > > >
>> > > > Rather than hang gliders and homebuilts, consider a Globalhawk that
>> > > > hits an airliner. Assuming no negligence on the part of the airliner
>> > > > crew, who do you go after? Do you go after the Air Force, Northrop
>> > > > Grumman, Raytheon, or somebody else? And of what are they likely to
>> > > > be found guilty?
>> >
>> > GlobalHawk drones do have human pilots. Although they are not on board, they
>> > are in control via a stellite link and can be thousands of miles away.
>> >
>> > .<http://www.aviationtoday.com/2017/03/16/day-life-us-air-force-drone-pilot/
>>
>> You are conflating Reaper and Globalhawk and totally missing the
>> point.
>
>Could you be more specific? Exactly what is wrong?

Reaper is a combat drone and is normally operated manually. We don't
let robots decided to shoot people yet. Globalhawk is a recon drone
and is normally autonomous. It has no weapons so shooting people is
not an issue. It can be operated manually and normally is in high
traffic areas for exactly the "what if it hits an airliner" reason,
but for most of its mission profile it is autonomous.

The article mentions Globalhawk in passing but then goes on to spend
the rest of its time discussing piloting Predator, which while still
in the inventory is ancestral to Reaper.

On Thu, 23 Nov 2017 23:00:51 -0800, OFWW <[email protected]>
wrote:

>So can we know return to the question or at the least, wood working?

Probably not

On Friday, November 24, 2017 at 9:11:22 AM UTC-5, J. Clarke wrote:
> On Thu, 23 Nov 2017 23:00:51 -0800, OFWW <[email protected]>
> wrote:
>=20
> >On Fri, 24 Nov 2017 00:37:20 -0500, J. Clarke
> ><[email protected]> wrote:
> >
> >>On Thu, 23 Nov 2017 20:52:09 -0800, OFWW <[email protected]>
> >>wrote:
> >>
> >>>On Thu, 23 Nov 2017 23:10:05 -0500, J. Clarke
> >>><[email protected]> wrote:
> >>>
> >>>>On Thu, 23 Nov 2017 18:44:05 -0800, OFWW <[email protected]>
> >>>>wrote:
> >>>>
> >>>>>On Thu, 23 Nov 2017 11:53:47 -0800 (PST), DerbyDad03
> >>>>><[email protected]> wrote:
> >>>>>
> >>>>>>On Thursday, November 23, 2017 at 11:40:13 AM UTC-5, OFWW wrote:
> >>>>>>> On Wed, 22 Nov 2017 12:36:05 -0800 (PST), DerbyDad03
> >>>>>>> <[email protected]> wrote:
> >>>>>>>=20
> >>>>>>> >On Wednesday, November 22, 2017 at 1:51:05 PM UTC-5, Ed Pawlowsk=
i wrote:
> >>>>>>> >> On 11/22/2017 1:20 PM, DerbyDad03 wrote:
> >>>>>>> >>=20
> >>>>>>> >> >=20
> >>>>>>> >> > Oh, well, no sense in waiting...
> >>>>>>> >> >=20
> >>>>>>> >> > 2nd scenario:
> >>>>>>> >> >=20
> >>>>>>> >> > 5 workers are standing on the railroad tracks. A train is he=
ading in their
> >>>>>>> >> > direction. They have no escape route. If the train continues=
down the tracks,
> >>>>>>> >> > it will most assuredly kill them all.
> >>>>>>> >> >=20
> >>>>>>> >> > You are standing on a bridge overlooking the tracks. Next to=
you is a fairly
> >>>>>>> >> > large person. We'll save you some trouble and let that perso=
n be a stranger.
> >>>>>>> >> >=20
> >>>>>>> >> > You have 2, and only 2, options. If you do nothing, all 5 wo=
rkers will
> >>>>>>> >> > be killed. If you push the stranger off the bridge, the trai=
n will kill
> >>>>>>> >> > him but be stopped before the 5 workers are killed. (Don't q=
uestion the
> >>>>>>> >> > physics, just accept the outcome.)
> >>>>>>> >> >=20
> >>>>>>> >> > Which option do you choose?
> >>>>>>> >> >=20
> >>>>>>> >>=20
> >>>>>>> >> I don't know. It was easy to pull the switch as there was a b=
it of=20
> >>>>>>> >> disconnect there. Now it is up close and you are doing the pu=
shing.=20
> >>>>>>> >> One alternative is to jump yourself, but I'd not do that. Don=
't think I=20
> >>>>>>> >> could push the guy either.
> >>>>>>> >>=20
> >>>>>>> >
> >>>>>>> >And there in lies the rub. The "disconnected" part.
> >>>>>>> >
> >>>>>>> >Now, as promised, let's bring this back to technology, AI and mo=
st=20
> >>>>>>> >certainly, people. Let's talk specifically about autonomous vehi=
cles,=20
> >>>>>>> >but please avoid the rabbit hole and realize that the concept ap=
plies
> >>>>>>> >to just about any where AI is used and people are involved. Auto=
nomus
> >>>>>>> >vehicles (AV) are just one example.
> >>>>>>> >
> >>>>>>> >Imagine it's X years from now and AV's are fairly common. Imagin=
e that an AV=20
> >>>>>>> >is traveling down the road, with its AI in complete control of t=
he vehicle.=20
> >>>>>>> >The driver is using one hand get a cup of coffee from the built-=
in Keurig=20
> >>>>>>> >machine and choosing a Pandora station with the other. He is com=
pletely=20
> >>>>>>> >oblivious to what's happening outside of his vehicle.=20
> >>>>>>> >
> >>>>>>> >Now imagine that a 4 year old runs out into the road. The AI use=
s all of the=20
> >>>>>>> >data at its disposal (speed, distance, weather conditions, tire =
pressure,=20
> >>>>>>> >etc.) and decides that it will not be able to stop in time. It c=
hecks the=20
> >>>>>>> >input from its 360=C2=B0 cameras. Can't go right because of the =
line of parked=20
> >>>>>>> >cars. They won't slow the vehicle enough to avoid hitting the ki=
d. Using=20
> >>>>>>> >facial recognition the AI determines that the mini-van on the le=
ft contains=20
> >>>>>>> >5 elderly people. If the AV swerves left, it will push the mini-=
van into=20
> >>>>>>> >oncoming traffic, directly into the path of a 18 wheeler. The AI=
communicates=20
> >>>>>>> >with the 18 wheeler's AI who responds and says "I have no place =
to go. If=20
> >>>>>>> >you push the van into my lane, I'm taking out a bunch of Grandma=
s and=20
> >>>>>>> >Grandpas."
> >>>>>>> >
> >>>>>>> >Now the AI has to make basically the same decision as in my firs=
t scenario:
> >>>>>>> >Kill 1 or kill 5. For the AI, it's as easy as it was for us, rig=
ht?=20
> >>>>>>> >
> >>>>>>> >"Bye Bye, kid. You should have stayed on the sidewalk."
> >>>>>>> >
> >>>>>>> >No emotion, right? Right, not once the AI is programmed, not onc=
e the initial
> >>>>>>> >AI rules have been written, not once the facial recognition data=
base has=20
> >>>>>>> >been built. The question is who wrote those rules? Who decided i=
t's OK to=20
> >>>>>>> >kill a young kid to save the lives of 5 rickety old folks? Oh wa=
it, maybe
> >>>>>>> >it's better to save the kid and let the old folks die. They've h=
ad a full
> >>>>>>> >life. Who wrote that rule? In other words, someone(s) have to de=
cide whose=20
> >>>>>>> >life is worth more than another's. They are essentially standing=
on a bridge=20
> >>>>>>> >deciding whether to push the guy or not. They have to write the =
rule. They=20
> >>>>>>> >are either going to kill the kid or push the car into the other =
lane.
> >>>>>>> >
> >>>>>>> >I, for one, don't think that I want to be sitting around that ta=
ble. Having=20
> >>>>>>> >to make the decisions would be one thing. Having to sit next to =
the person
> >>>>>>> >that would push the guy off the bridge with a gleam in his eye w=
ould be a
> >>>>>>> >totally different story.=20
> >>>>>>>=20
> >>>>>>> I reconsidered my thoughts on this one as well.
> >>>>>>>=20
> >>>>>>> The AV should do as it was designed to do, to the best of its
> >>>>>>> capabilities. Staying in the lane when there is no option to swer=
ve
> >>>>>>> safely.
> >>>>>>>=20
> >>>>>>> There is already a legal reason for that, that being that the swe=
rving
> >>>>>>> driver assumes all the damages that incur from his action, includ=
ing
> >>>>>>> manslaughter.
> >>>>>>
> >>>>>>So in the following brake failure scenario, if the AV stays in lane=
and=20
> >>>>>>kills the four "highly rated" pedestrians there are no charges, but=
if=20
> >>>>>>it changes lanes and takes out the 4 slugs, jail time may ensue.
> >>>>>>
> >>>>>>http://static6.businessinsider.com/image/58653ba0ee14b61b008b5aea-8=
00
> >>>>>>
> >>>>>>Interesting.
> >>>>>
> >>>>>Yes, and I've been warned that by my taking evasive action I could
> >>>>>cause someone else to respond likewise and that I would he held
> >>>>>accountable for what happened.
> >>>>
> >>>>I find the assumption that a fatality involving a robot car would lea=
d
> >>>>to someone being jailed to be amusing. The people who assert this
> >>>>never identify the statute under which someone would be jailed or who=
,
> >>>>precisely this someone might be. They seem to assume that because a
> >>>>human driving a car could be jailed for vehicular homicide or crimina=
l
> >>>>negligence or some such, it is automatic that someone else would be
> >>>>jailed for the same offense--the trouble is that the car is legally a=
n
> >>>>inanimate object and we don't put inanimate objects in jail. So it
> >>>>gets down to proving that the occupant is negligent, which is a hard
> >>>>sell given that the government allowed the car to be licensed with th=
e
> >>>>understanding that it would not be controlled by the occupant, or
> >>>>proving that the engineering team responsible for developing it was
> >>>>negligent, which given that they can show the logic the thing used an=
d
> >>>>no doubt provide legal justification for the decision it made, will b=
e
> >>>>another tall order. So who goes to jail?
> >>>>
> >>>
> >>>You've taken it to the next level, into the real word scenario and out
> >>>of the programming stage.
> >>>
> >>>Personally I would assume that anything designed would have to
> >>>co-exist with real world laws and responsibilities. Even the final
> >>>owner could be held responsible. See the laws regarding experimental
> >>>aircraft, hang gliders, etc.
> >>
> >>Experimental aircraft and hang gliders are controlled by a human. If
> >>they are involved in a fatl accident, the operator gets scrutinized.
> >>An autonomous car is not under human control, it is its own operator,
> >>the occupant is a passenger.
> >>
> >>We don't have "real world law" governing fatalities involving
> >>autonomous vehicles. The engineering would, initially (I hope) be
> >>based on existing case law involving human drivers and what the courts
> >>held that they should or should not have done in particular
> >>situations. But there won't be any actual law until either the
> >>legislatures write statutes or the courts issue rulings, and the
> >>latter won't happen until there are such vehicles in service in
> >>sufficient quantity to generate cases.
> >>
> >>Rather than hang gliders and homebuilts, consider a Globalhawk that
> >>hits an airliner. Assuming no negligence on the part of the airliner
> >>crew, who do you go after? Do you go after the Air Force, Northrop
> >>Grumman, Raytheon, or somebody else? And of what are they likely to
> >>be found guilty?
> >>
> >>>But we should be sticking to this hypothetical example given us.
> >>
> >>It was suggested that someone would go to jail. I still want to know
> >>who and what crime they committed.
> >
> >The person who did not stay in their own lane, and ended up committing
> >involuntary manslaughter.
>=20
> Are you arguing that an autonomous vehicle is a "person"? You
> really don't seem to grasp the concept. Rather than a car with an
> occupant, make it a car, say a robot taxicab, that is going somewhere
> or other unoccupied.
>=20
> >In the case you bring up the AV can be currently over ridden at
> >anytime by the occupant. There are already AV vehicles operating on
> >the streets.
>=20
> In what case that I bring up? Globalhawk doesn't _have_ an occupant.
> (when people use words with which you are unfamiliar, you should at
> least Google those words before opining). There are very few
> autonomous vehicles and currently they are for the most part operated
> with a safety driver, but that is not anybody's long-term plan. Google
> already has at least one demonstrator with no steering wheel or pedals
> and Uber is planning on using driverless cars in their ride sharing
> service--ultimately those would also have no controls accessible to
> the passenger.
>=20
> >Regarding your "whose at fault" scenario, just look at the court cases
> >against gun makers, as if guns kill people.
>=20
> I have not introduced a "who's at fault scenariao". I have asked what
> law would be violated and who would be jailed. "At fault" decides who
> pays damages, not who goes to jail. I am not discussing damages, I am
> discussing JAIL. You do know what a jail is, do you not?
>=20
> >So can we know return to the question or at the least, wood working?
>=20
> You're the one who started feeding the troll.

...and then you joined the meal.

On Thu, 23 Nov 2017 23:00:51 -0800, OFWW <[email protected]>
wrote:

>On Fri, 24 Nov 2017 00:37:20 -0500, J. Clarke
><[email protected]> wrote:
>
>>On Thu, 23 Nov 2017 20:52:09 -0800, OFWW <[email protected]>
>>wrote:
>>
>>>On Thu, 23 Nov 2017 23:10:05 -0500, J. Clarke
>>><[email protected]> wrote:
>>>
>>>>On Thu, 23 Nov 2017 18:44:05 -0800, OFWW <[email protected]>
>>>>wrote:
>>>>
>>>>>On Thu, 23 Nov 2017 11:53:47 -0800 (PST), DerbyDad03
>>>>><[email protected]> wrote:
>>>>>
>>>>>>On Thursday, November 23, 2017 at 11:40:13 AM UTC-5, OFWW wrote:
>>>>>>> On Wed, 22 Nov 2017 12:36:05 -0800 (PST), DerbyDad03
>>>>>>> <[email protected]> wrote:
>>>>>>>
>>>>>>> >On Wednesday, November 22, 2017 at 1:51:05 PM UTC-5, Ed Pawlowski wrote:
>>>>>>> >> On 11/22/2017 1:20 PM, DerbyDad03 wrote:
>>>>>>> >>
>>>>>>> >> >
>>>>>>> >> > Oh, well, no sense in waiting...
>>>>>>> >> >
>>>>>>> >> > 2nd scenario:
>>>>>>> >> >
>>>>>>> >> > 5 workers are standing on the railroad tracks. A train is heading in their
>>>>>>> >> > direction. They have no escape route. If the train continues down the tracks,
>>>>>>> >> > it will most assuredly kill them all.
>>>>>>> >> >
>>>>>>> >> > You are standing on a bridge overlooking the tracks. Next to you is a fairly
>>>>>>> >> > large person. We'll save you some trouble and let that person be a stranger.
>>>>>>> >> >
>>>>>>> >> > You have 2, and only 2, options. If you do nothing, all 5 workers will
>>>>>>> >> > be killed. If you push the stranger off the bridge, the train will kill
>>>>>>> >> > him but be stopped before the 5 workers are killed. (Don't question the
>>>>>>> >> > physics, just accept the outcome.)
>>>>>>> >> >
>>>>>>> >> > Which option do you choose?
>>>>>>> >> >
>>>>>>> >>
>>>>>>> >> I don't know. It was easy to pull the switch as there was a bit of
>>>>>>> >> disconnect there. Now it is up close and you are doing the pushing.
>>>>>>> >> One alternative is to jump yourself, but I'd not do that. Don't think I
>>>>>>> >> could push the guy either.
>>>>>>> >>
>>>>>>> >
>>>>>>> >And there in lies the rub. The "disconnected" part.
>>>>>>> >
>>>>>>> >Now, as promised, let's bring this back to technology, AI and most
>>>>>>> >certainly, people. Let's talk specifically about autonomous vehicles,
>>>>>>> >but please avoid the rabbit hole and realize that the concept applies
>>>>>>> >to just about any where AI is used and people are involved. Autonomus
>>>>>>> >vehicles (AV) are just one example.
>>>>>>> >
>>>>>>> >Imagine it's X years from now and AV's are fairly common. Imagine that an AV
>>>>>>> >is traveling down the road, with its AI in complete control of the vehicle.
>>>>>>> >The driver is using one hand get a cup of coffee from the built-in Keurig
>>>>>>> >machine and choosing a Pandora station with the other. He is completely
>>>>>>> >oblivious to what's happening outside of his vehicle.
>>>>>>> >
>>>>>>> >Now imagine that a 4 year old runs out into the road. The AI uses all of the
>>>>>>> >data at its disposal (speed, distance, weather conditions, tire pressure,
>>>>>>> >etc.) and decides that it will not be able to stop in time. It checks the
>>>>>>> >input from its 360° cameras. Can't go right because of the line of parked
>>>>>>> >cars. They won't slow the vehicle enough to avoid hitting the kid. Using
>>>>>>> >facial recognition the AI determines that the mini-van on the left contains
>>>>>>> >5 elderly people. If the AV swerves left, it will push the mini-van into
>>>>>>> >oncoming traffic, directly into the path of a 18 wheeler. The AI communicates
>>>>>>> >with the 18 wheeler's AI who responds and says "I have no place to go. If
>>>>>>> >you push the van into my lane, I'm taking out a bunch of Grandmas and
>>>>>>> >Grandpas."
>>>>>>> >
>>>>>>> >Now the AI has to make basically the same decision as in my first scenario:
>>>>>>> >Kill 1 or kill 5. For the AI, it's as easy as it was for us, right?
>>>>>>> >
>>>>>>> >"Bye Bye, kid. You should have stayed on the sidewalk."
>>>>>>> >
>>>>>>> >No emotion, right? Right, not once the AI is programmed, not once the initial
>>>>>>> >AI rules have been written, not once the facial recognition database has
>>>>>>> >been built. The question is who wrote those rules? Who decided it's OK to
>>>>>>> >kill a young kid to save the lives of 5 rickety old folks? Oh wait, maybe
>>>>>>> >it's better to save the kid and let the old folks die. They've had a full
>>>>>>> >life. Who wrote that rule? In other words, someone(s) have to decide whose
>>>>>>> >life is worth more than another's. They are essentially standing on a bridge
>>>>>>> >deciding whether to push the guy or not. They have to write the rule. They
>>>>>>> >are either going to kill the kid or push the car into the other lane.
>>>>>>> >
>>>>>>> >I, for one, don't think that I want to be sitting around that table. Having
>>>>>>> >to make the decisions would be one thing. Having to sit next to the person
>>>>>>> >that would push the guy off the bridge with a gleam in his eye would be a
>>>>>>> >totally different story.
>>>>>>>
>>>>>>> I reconsidered my thoughts on this one as well.
>>>>>>>
>>>>>>> The AV should do as it was designed to do, to the best of its
>>>>>>> capabilities. Staying in the lane when there is no option to swerve
>>>>>>> safely.
>>>>>>>
>>>>>>> There is already a legal reason for that, that being that the swerving
>>>>>>> driver assumes all the damages that incur from his action, including
>>>>>>> manslaughter.
>>>>>>
>>>>>>So in the following brake failure scenario, if the AV stays in lane and
>>>>>>kills the four "highly rated" pedestrians there are no charges, but if
>>>>>>it changes lanes and takes out the 4 slugs, jail time may ensue.
>>>>>>
>>>>>>http://static6.businessinsider.com/image/58653ba0ee14b61b008b5aea-800
>>>>>>
>>>>>>Interesting.
>>>>>
>>>>>Yes, and I've been warned that by my taking evasive action I could
>>>>>cause someone else to respond likewise and that I would he held
>>>>>accountable for what happened.
>>>>
>>>>I find the assumption that a fatality involving a robot car would lead
>>>>to someone being jailed to be amusing. The people who assert this
>>>>never identify the statute under which someone would be jailed or who,
>>>>precisely this someone might be. They seem to assume that because a
>>>>human driving a car could be jailed for vehicular homicide or criminal
>>>>negligence or some such, it is automatic that someone else would be
>>>>jailed for the same offense--the trouble is that the car is legally an
>>>>inanimate object and we don't put inanimate objects in jail. So it
>>>>gets down to proving that the occupant is negligent, which is a hard
>>>>sell given that the government allowed the car to be licensed with the
>>>>understanding that it would not be controlled by the occupant, or
>>>>proving that the engineering team responsible for developing it was
>>>>negligent, which given that they can show the logic the thing used and
>>>>no doubt provide legal justification for the decision it made, will be
>>>>another tall order. So who goes to jail?
>>>>
>>>
>>>You've taken it to the next level, into the real word scenario and out
>>>of the programming stage.
>>>
>>>Personally I would assume that anything designed would have to
>>>co-exist with real world laws and responsibilities. Even the final
>>>owner could be held responsible. See the laws regarding experimental
>>>aircraft, hang gliders, etc.
>>
>>Experimental aircraft and hang gliders are controlled by a human. If
>>they are involved in a fatl accident, the operator gets scrutinized.
>>An autonomous car is not under human control, it is its own operator,
>>the occupant is a passenger.
>>
>>We don't have "real world law" governing fatalities involving
>>autonomous vehicles. The engineering would, initially (I hope) be
>>based on existing case law involving human drivers and what the courts
>>held that they should or should not have done in particular
>>situations. But there won't be any actual law until either the
>>legislatures write statutes or the courts issue rulings, and the
>>latter won't happen until there are such vehicles in service in
>>sufficient quantity to generate cases.
>>
>>Rather than hang gliders and homebuilts, consider a Globalhawk that
>>hits an airliner. Assuming no negligence on the part of the airliner
>>crew, who do you go after? Do you go after the Air Force, Northrop
>>Grumman, Raytheon, or somebody else? And of what are they likely to
>>be found guilty?
>>
>>>But we should be sticking to this hypothetical example given us.
>>
>>It was suggested that someone would go to jail. I still want to know
>>who and what crime they committed.
>
>The person who did not stay in their own lane, and ended up committing
>involuntary manslaughter.

Are you arguing that an autonomous vehicle is a "person"? You
really don't seem to grasp the concept. Rather than a car with an
occupant, make it a car, say a robot taxicab, that is going somewhere
or other unoccupied.

>In the case you bring up the AV can be currently over ridden at
>anytime by the occupant. There are already AV vehicles operating on
>the streets.

In what case that I bring up? Globalhawk doesn't _have_ an occupant.
(when people use words with which you are unfamiliar, you should at
least Google those words before opining). There are very few
autonomous vehicles and currently they are for the most part operated
with a safety driver, but that is not anybody's long-term plan. Google
already has at least one demonstrator with no steering wheel or pedals
and Uber is planning on using driverless cars in their ride sharing
service--ultimately those would also have no controls accessible to
the passenger.

>Regarding your "whose at fault" scenario, just look at the court cases
>against gun makers, as if guns kill people.

I have not introduced a "who's at fault scenariao". I have asked what
law would be violated and who would be jailed. "At fault" decides who
pays damages, not who goes to jail. I am not discussing damages, I am
discussing JAIL. You do know what a jail is, do you not?

>So can we know return to the question or at the least, wood working?

You're the one who started feeding the troll.