AI & Origins of Siri

MrAl

Joined Jun 17, 2014
6,608
Hi,

I think the test was that a computer and a human would be hiding from the second human, and the second human can ask any questions it wants to ask, and if the second human can not tell if it is a computer or human answering then the computer has reached the point of true imitation of the human.

If you are interested in AI and have not yet studied automated reasoning, that will blow your mind :)
That's like the machine part of reasoning, which goes above and beyond the way humans think.
 

nsaspook

Joined Aug 27, 2009
6,473
The true imitation of the humans will be the creation of strong artificial ignorance. How does a computer know when not to answer a question that's beyond normal human calculation or recall ability that would reveal it's true identity or motives? Intelligence is far over-rated as a human trait because it's very easy to fool. The ability to cheat, manipulate and lie conveniently is what made HAL9000 believable as an AI

https://www.redteamsecure.com/5-effective-social-engineering-elicitation-techniques/
 

MrAl

Joined Jun 17, 2014
6,608
Hi,

Yeah we have all seen and heard computers on TV being tricked into entering an endless loop after being given a question that presents a problem. An example (a little funny too) is from South Park, where the boys trick the "Funny Bot" robot that wants to destroy the world as the ultimate joke into an endless loop by awarding it a physical "Comedy Award" in the form of a small statue. Upon receiving the award, Funny Bot went into an endless loop because it reasoned that comedy was not serious but if it was getting an award then that means it had to take comedy seriously, so it kept repeating that to itself and eventually blew out some circuits and shut down.

There are some much better examples though that are not funny at all. They represent serious problems in reasoning. The most interesting one i think is where the subject of the reasoning argument is part of the whole, and the argument makes statements about the subject as a whole. I'd have to look up this one though as i dont remember the exact wording. One of them is when a question is asked of a computer program and it answers but at the same time that commits to a certain way of reasoning and so another question answer will contradict that one and so the computer cant answer it because it can not allow a contradiction. It's like how do you believe a liar when they say they are not lying :)
 

Thread Starter

djsfantasi

Joined Apr 11, 2010
5,683
T
Hi,

Yeah we have all seen and heard computers on TV being tricked into entering an endless loop after being given a question that presents a problem. An example (a little funny too) is from South Park, where the boys trick the "Funny Bot" robot that wants to destroy the world as the ultimate joke into an endless loop by awarding it a physical "Comedy Award" in the form of a small statue. Upon receiving the award, Funny Bot went into an endless loop because it reasoned that comedy was not serious but if it was getting an award then that means it had to take comedy seriously, so it kept repeating that to itself and eventually blew out some circuits and shut down.

There are some much better examples though that are not funny at all. They represent serious problems in reasoning. The most interesting one i think is where the subject of the reasoning argument is part of the whole, and the argument makes statements about the subject as a whole. I'd have to look up this one though as i dont remember the exact wording. One of them is when a question is asked of a computer program and it answers but at the same time that commits to a certain way of reasoning and so another question answer will contradict that one and so the computer cant answer it because it can not allow a contradiction. It's like how do you believe a liar when they say they are not lying :)
This is the basis of many of Isaac Asimov's stories in the I, Robot series...
 
Top