|
|
mark merrens wrote: Should Robot Cars Be Programmed To Kill You If It Will Save More Lives?
No. No AI bot should ever have the ability to judge the value life. How can it? It has no concept of it. To think people actually have to ask this question.
Jeremy Falcon
|
|
|
|
|
So, given that the bot is able to predict the outcome of the accident and knowing that only 2 rather than, say, 6 people will die it should not take that choice?
Jeremy Falcon wrote: No AI bot should ever have the ability to judge the value life. How can it? It has no concept of it.
It is because it is acting without emotion that it can make this decision. It is you humans who are incapable of doing that.
Oh...
"If you think it's expensive to hire a professional to do the job, wait until you hire an amateur." Red Adair.
Those who seek perfection will only find imperfection
nils illegitimus carborundum
me, me, me
me, in pictures
|
|
|
|
|
mark merrens wrote: So, given that the bot is able to predict the outcome of the accident and knowing that only 2 rather than, say, 6 people will die it should not take that choice?
You cannot make a judgment call on that like it's a simple logical algorithm in a program. What if the person to die was your daughter, who's also pregnant, and her husband? And the people living were 6 old people that were murderers and on their way to kill more people? Oh sure, then we could have the cars cop a feel for pregnant chicks every time you start the car and require old people to sign a waver to kiss their arse good bye. But where does it stop, just how far down the "lets not have to think for ourselves" rabbit hole does one have to go? Just because technology says we can.
mark merrens wrote: It is because it is acting without emotion that it can make this decision. It is you humans who are incapable of doing that.
Har har. Seriously though, emotion is what makes life worth living. It's what makes being human fun. Oh wait that's an emotion. I just want to be happy. Oh wait... damn emotions getting in the way.
Einstein was right if this question even has to be asked.
Jeremy Falcon
|
|
|
|
|
Jeremy Falcon wrote: You cannot make a judgment call on that like it's a simple logical algorithm in a program.
I think that's exactly what you can do. The bot will not make judgments about the people only about the outcome.
In any case, if all cars were guided by bots this is unlikely to happen other than by freak, unpredictable accident.
"If you think it's expensive to hire a professional to do the job, wait until you hire an amateur." Red Adair.
Those who seek perfection will only find imperfection
nils illegitimus carborundum
me, me, me
me, in pictures
|
|
|
|
|
mark merrens wrote: I think that's exactly what you can do. The bot will not make judgments about the people only about the outcome.
Yeah totally. Let's give robots the power of God. That shouldn't cause any problems. 
Jeremy Falcon
|
|
|
|
|
It's not the power of god, it's the power of reason and making logical judgments based on probability.
"If you think it's expensive to hire a professional to do the job, wait until you hire an amateur." Red Adair.
Those who seek perfection will only find imperfection
nils illegitimus carborundum
me, me, me
me, in pictures
|
|
|
|
|
Yeah... go right on and believe that.
Jeremy Falcon
|
|
|
|
|
mark merrens wrote: making logical judgments based on probability
Yeah... I prefer having some odds than no odds at all. So... slam them.
To alcohol! The cause of, and solution to, all of life's problems - Homer Simpson
----
Our heads are round so our thoughts can change direction - Francis Picabia
|
|
|
|
|
Jeremy Falcon wrote: Let's give robots the power of God. That shouldn't cause any problems.
It would be interesting to see if robots whose behavior was driven by neural-nets based on the behavior of human beings who claim to act/speak for/in-the-name-of "gods" were equally blood-thirsty and came up with pogroms, ethnic cleansings, and atrocities equal to said human beings.
“I speak in a poem of the ancient food of heroes: humiliation, unhappiness, discord. Those things are given to us to transform, so that we may make from the miserable circumstances of our lives things that are eternal, or aspire to be so.” Jorge Luis Borges
|
|
|
|
|
BillWoodruff wrote: were equally blood-thirsty and came up with pogroms, ethnic cleansings, and atrocities equal to said human beings.
For being human yourself, you really have a low opinion of them. I happen to quite like humans. I think I'll stay one.
Jeremy Falcon
|
|
|
|
|
mark merrens wrote: In any case, if all cars were guided by bots this is unlikely to happen
Unless VB programmers are involved.
Or, a mismatch between metric and US measurement systems.
There are only 10 types of people in the world, those who understand binary and those who don't.
|
|
|
|
|
"If you think it's expensive to hire a professional to do the job, wait until you hire an amateur." Red Adair.
Those who seek perfection will only find imperfection
nils illegitimus carborundum
me, me, me
me, in pictures
|
|
|
|
|
Jeremy Falcon wrote: And the people living were 6 old people that were murderers and on their way to kill more people?
In that case lets just program the robotic car of these to fataly crash, removing them from other cars equations 
Seriously, though: how do you know this is the case? And if you know, why can't the car? Why can't that other people's car? Why can't that other people's car decide and ... oh well, back to my initial statement again 
|
|
|
|
|
This will just bring on more car hacking. Use a car key to send an encoded signal which overflows a correct-key-match buffer and tells the car it really needs to kill all its occupants. National Security and hired assassinations made easy.
|
|
|
|
|
The premise already is that the robotic car is programmed to kill its occupants under certain conditions (presumably to minimize the overall loss). I merely suggested additional conditions. And, yes, however these conditions are programmed, any software system can and will be hacked and abused.
The question is, how much damage will be incurred through abuse, manipulation, or just honest software errors, compared to the damage these systems may avert...
|
|
|
|
|
mark merrens wrote: It is you humans
Aha! I knew it.
There are only 10 types of people in the world, those who understand binary and those who don't.
|
|
|
|
|
Oops! My positronic pathways must have malfunctioned for a nano-second.
"If you think it's expensive to hire a professional to do the job, wait until you hire an amateur." Red Adair.
Those who seek perfection will only find imperfection
nils illegitimus carborundum
me, me, me
me, in pictures
|
|
|
|
|
mark merrens wrote: for a nano-second.
Which is like 1 human hour.
There are only 10 types of people in the world, those who understand binary and those who don't.
|
|
|
|
|
You have programed the bot to make the value judgement that killing 2 people is always better than killing 6.
|
|
|
|
|
It's not a value judgment: it's simple arithmetic.
"If you think it's expensive to hire a professional to do the job, wait until you hire an amateur." Red Adair.
Those who seek perfection will only find imperfection
nils illegitimus carborundum
me, me, me
me, in pictures
|
|
|
|
|
Not making a choice is a choice as well.
But what if the computer has two options:
-Keep driving ahead and kill x pedestrians. ('do nothing')
-Steer the car into the nearest tree and kill y passengers.
All other possibilities have been evaluated and determined to be physically impossible.
(speed too high, braking distance too short, trees on both sides of the road, etc)
What should the computer do when there is no 'do nothing'?
If the decision of who is killed cannot be made by a computer then it must be escalated to a human.
But to which human?
-The passengers?
-The pedestrians?
Both have a personal interest in the decision so neither can be trusted to be fair.
Maybe the decision should be deferred to an impartial referee?
The computer could warn a government official, present him with all relevant data and then let him make a choice.
Or make the decision through a democratic process.
Ask a large number of responsible citizens what action hould be taken and then take the most popular course of action.
This can be done with modern technology.
Just get a notification on your smartphone with a small animation of each option and then tap the one you favor.
You could even disguise it as a game.
|
|
|
|
|
jeroen1304 wrote: But what if the computer has two options:
-Keep driving ahead and kill x pedestrians. ('do nothing')
-Steer the car into the nearest tree and kill y passengers.
Exactly this was the question in the article ...
Quote: jeroen1304 wrote: All other possibilities have been evaluated and determined to be physically impossible.
(speed too high, braking distance too short, trees on both sides of the road, etc)
jeroen1304 wrote: make the decision through a democratic process.
Ask a large number of responsible citizens what action hould be taken and then take the most popular
course of action.
You cannot take this route, 'cause there is no time for it. The decision has to be made in fractions of the next second.
|
|
|
|
|
I think that's a fascinating scenario to think about, Mark. Consider the robot-in-the-car detects loss of consciousness in the driver somehow and is able to evaluate, given the flow of traffic, that any sudden stop will result in a multi-car pile-up with major loss of life while it is also able to conclude that a sudden sharp turn will take the vehicle off the roadway, but almost certainly kill the occupant.
Medical personnel in war, given an overflow of casualties, make rapid decisions (triage) about who gets treatment priority based on intuitive mortality assessments as well as, of course, whatever medical stats they can get. It would be interesting, to me, to know to what extent the current state-of-the-art triage strategies in war and natural disasters are using computer programs to assist evaluation.
Equally frightening is the idea of a "loyal" robot programmed to put the preservation of its owner above everyone/everything else. I observe that my mind associates the terms "loyal robot" with the typical spin-minions and henchmen/women of ... politicians.
cheers, Bill
“I speak in a poem of the ancient food of heroes: humiliation, unhappiness, discord. Those things are given to us to transform, so that we may make from the miserable circumstances of our lives things that are eternal, or aspire to be so.” Jorge Luis Borges
modified 13-May-14 18:58pm.
|
|
|
|
|
BillWoodruff wrote: I think that's a fascinating scenario to think about, Mark. Consider the robot-in-the-car detects loss of consciousness in the driver somehow and is able to evaluate, given the flow of traffic, that any sudden stop will result in a multi-car pile-up with major loss of life while it is also able to conclude that a sudden sharp turn will take the vehicle off the roadway, but almost certainly kill the occupant.
Sounds rather unrealistic to me:
1. If the car is robot-controlled to start with, why can't it just go on driving?
2. If stopping your car could potentially cause lifes, what the hell were the other drivers/robot cars thinking?
3. If the other cars are also robot-controlled, why can't they collaborate to ensure a safe mutual slowdown?
4. Can't think of any reason why a sharp turn would be less dangerous to the rest of the traffic
BillWoodruff wrote: Equally frightening is the idea of a "loyal" robot programmed to put the preservation of its owner above everyone/everything else. I observe that my mind associates the terms "loyal robot" with the typical spin-minions and henchmen/women of ... politicians.
That could indeed be a problem, and car makers could in fact promote cars with 'improved survivability' for those who are willing to shell out the cash. Politicians could try to prevent that, but, realistically, by the time they can agree on a workable legislation the market will already be brimming with such discriminating cars that are hard to tone down or remove.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
