If you thought Skynet was just a story why is google building an AI kill switch
-
@dafyre said in If you thought Skynet was just a story why is google building an AI kill switch:
@thwr said in If you thought Skynet was just a story why is google building an AI kill switch:
https://en.wikipedia.org/wiki/Three_Laws_of_Robotics
- A robot may not injure a human being or, through inaction, allow a human being to come to harm.
- A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
- A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.
Rule #3 is a real problem. A little mistake in the implementation and ... well, let's hope there's a Connor family around. A kill switch isn't the dumbest thing to implement.
Why should a robot protect its own existence? Save the human, sacrifice yourself, lol.
To save the money it takes to make it but it's just a priority list. Humans come first. If they aren't in danger then save yourself.
Isn't this from iRobot?
-
@wirestyle22 It is from iRobot.
-
@dafyre That's something scientists are thinking about ever since 1942. Moral and ethics in robotics / AI. Been reading an article some months ago about self-piloting cars.
What should the machine do in the following case:
A familiy is near the road. An old lady is just crossing the same road. The driver has a heart attack in this very moment and can't control the car. The speed is too high to come to a stop. What is the car supposed to do? Kill the family? Kill the old lady? -
@DustinB3403 said in If you thought Skynet was just a story why is google building an AI kill switch:
@wirestyle22 It is from iRobot.
It's way older..."The Three Laws of Robotics (often shortened to The Three Laws or Three Laws, also known as Asimov's Laws) are a set of rules devised by the science fiction author Isaac Asimov." https://en.wikipedia.org/wiki/Three_Laws_of_Robotics
-
@thwr said in If you thought Skynet was just a story why is google building an AI kill switch:
What should the machine do in the following case:
A familiy is near the road. An old lady is just crossing the same road. The driver has a heart attack in this very moment and can't control the car. The speed is too high to come to a stop. What is the car supposed to do? Kill the family? Kill the old lady?What control does the machine have in that question?
-
@scottalanmiller Full control: speed, brakes, stearing, traction control, anti-skid braking system...
The article was in German I'm afraid, written by a German professor researching robotics and AI.
-
@scottalanmiller said in If you thought Skynet was just a story why is google building an AI kill switch:
@thwr said in If you thought Skynet was just a story why is google building an AI kill switch:
What should the machine do in the following case:
A familiy is near the road. An old lady is just crossing the same road. The driver has a heart attack in this very moment and can't control the car. The speed is too high to come to a stop. What is the car supposed to do? Kill the family? Kill the old lady?What control does the machine have in that question?
I would think the machine would have to make the decision to take out the elderly person. It's one life versus many lives
-
@DustinB3403 said in If you thought Skynet was just a story why is google building an AI kill switch:
@scottalanmiller said in If you thought Skynet was just a story why is google building an AI kill switch:
@thwr said in If you thought Skynet was just a story why is google building an AI kill switch:
What should the machine do in the following case:
A familiy is near the road. An old lady is just crossing the same road. The driver has a heart attack in this very moment and can't control the car. The speed is too high to come to a stop. What is the car supposed to do? Kill the family? Kill the old lady?What control does the machine have in that question?
I would think the machine would have to make the decision to take out the elderly person. It's one life versus many lives
In this case, a machine is deciding upon the value of a single life, and that's terribly dangerous. What if the old lady was like Einstein and the family had no job at all?
-
Don't get me wrong, if the AI was able to chose, hopefully it would chose to "not take any life" but if the only options were
- Kill 1 person
- Kill multiple people
It would always chose the lesser death toll. I would hope.
-
@thwr said in If you thought Skynet was just a story why is google building an AI kill switch:
@scottalanmiller Full control: speed, brakes, stearing, traction control, anti-skid braking system...
The article was in German I'm afraid, written by a German professor researching robotics and AI.
What vehicle is it controlling?
-
Where I am confused...
There is a driver with a heart attack in the car. there is an old lady in the road. Where does the machine or the family come into the picture?
-
@scottalanmiller said in If you thought Skynet was just a story why is google building an AI kill switch:
@thwr said in If you thought Skynet was just a story why is google building an AI kill switch:
@scottalanmiller Full control: speed, brakes, stearing, traction control, anti-skid braking system...
The article was in German I'm afraid, written by a German professor researching robotics and AI.
What vehicle is it controlling?
The machine is a self-piloting car, for example a modern BMW 5 series or a SUV.
-
@scottalanmiller Old lady on left side of road, the family is on the other side.
The AI has to stop the vehicle, but the only way to do it is to hit 1 person or a family.
Who should the AI kill?
-
@scottalanmiller said in If you thought Skynet was just a story why is google building an AI kill switch:
Where I am confused...
There is a driver with a heart attack in the car. there is an old lady in the road. Where does the machine or the family come into the picture?
It's about the machine having no choice. Someone will have an accident, either the family or the lady. The driver can't step in because he is suffering from the heart attack.
-
@thwr said in If you thought Skynet was just a story why is google building an AI kill switch:
@scottalanmiller said in If you thought Skynet was just a story why is google building an AI kill switch:
@thwr said in If you thought Skynet was just a story why is google building an AI kill switch:
@scottalanmiller Full control: speed, brakes, stearing, traction control, anti-skid braking system...
The article was in German I'm afraid, written by a German professor researching robotics and AI.
What vehicle is it controlling?
The machine is a self-piloting car, for example a modern BMW 5 series or a SUV.
And the machine had a heart attack?
-
The driver or the car (machine), not the machine
-
@DustinB3403 said in If you thought Skynet was just a story why is google building an AI kill switch:
@scottalanmiller Old lady on left side of road, the family is on the other side.
The AI has to stop the vehicle, but the only way to do it is to hit 1 person or a family.
Who should the AI kill?
Well it certainly shouldn't go off roading just to hit people.
-
There is no wrong or right in my opinion. It's an ethical dilemma, but very well worth discussing it.
-
@thwr said in If you thought Skynet was just a story why is google building an AI kill switch:
The driver or the car (machine), not the machine
I'm confused. There is a self driving car but there is a driver? Is it self piloting but not being self piloted? Why is a person driving a self piloting car?
-
@scottalanmiller said in If you thought Skynet was just a story why is google building an AI kill switch:
@thwr said in If you thought Skynet was just a story why is google building an AI kill switch:
@scottalanmiller said in If you thought Skynet was just a story why is google building an AI kill switch:
@thwr said in If you thought Skynet was just a story why is google building an AI kill switch:
@scottalanmiller Full control: speed, brakes, stearing, traction control, anti-skid braking system...
The article was in German I'm afraid, written by a German professor researching robotics and AI.
What vehicle is it controlling?
The machine is a self-piloting car, for example a modern BMW 5 series or a SUV.
And the machine had a heart attack?
No the car has a driver, but is capable of self-piloting. The car was in "human pilot mode" and the driver has a heart attack. The car senses "I'm not under the control of the human any more, time to take over"
But I'm out of control the only way to stop is to crash either to the left or right. Which should the AI chose?
