115 réponses à ce sujet

[Obadiah]

Going to see Age of Ultron today. I imagine that is how it would happen. Peace no matter what.

I wouldn't look to Ultron for any especially deep insights into AI.

[Draining Dragon]

Maybe we should start with the natural kind first.

[mybudgee]

[Manc4life7]

I totally get the AGI -> ASI thing. That kinda makes sense to me. One problem I can anticipate is that a being (whatever it's form) doesn't put the same value to human life as we do. Basically, just like we don't put a lot of value to the life of a mosquito.

 

What I always find amusing though is that everyone seems to expect that these ASIs would want to kill us. I mean, will kill a mosquito if it bugs me but I won't go hunting for one.

I wouldn't have a problem, being the favorite pet human of my master super-intelligent robot. Maybe it'll even scratch my belly.

 

That's basically the main premise of the discussion in the book I posted back on the first page - the dangers posed by an AGI or ASI without the proper regard for human life.  It turns out most people working on AI don't seem to put much thought into making sure any AI they create will be a "safe" AI, which is the author's main concern.  And you are right, far too many people who think AI is a horrible idea do so because they fear it will be evil - they anthropomorphize it - like we might actually get Skynet or the starchild.

 

AGI & ASI won't give two craps about ruling the world or wiping out humanity - it will go about achieving it's goals, whatever they may be, and that's it.  The danger lies in how it goes about achieving it's goals.  The most common example of a "bad" ASI would be one that starts as an AGI, built for some mundane task (such as producing handwritten Thank You cards, indistinguishable from one written by a human), that then goes through a hard takeoff, achieves ASI before anyone realizes it, and then goes about consuming everything on the face of the Earth via nano-assemblers invented in order to more efficiently complete it's task via blunt force repetition of it's goal.

 

Sounds freaky and somewhat silly, but entirely plausible.  From most of what I've read recently, many people way smarter than me think we finally get to useable and efficient nano-tech via AI inventing it for us.

[Jeremiah12LGeek]

I've always thought the phrase should be "synthetic intelligence," since that would be more accurate. "Artificial" is more like what we have now with AI, which is the appearance of intelligence, rather than actual intelligence.

[mybudgee]

I've always thought the phrase should be "synthetic intelligence," since that would be more accurate. "Artificial" is more like what we have now with AI, which is the appearance of intelligence, rather than actual intelligence.

That or "Jersey Shore"

[Commander Rpg]

http://en.wikipedia....ki/Chinese_room

[mybudgee]

Wow... Not really any Ultron comments. what's up with that?

[The Devlish Redhead]

Sign me up for a real life EDI when this happens.

 

She could help me out gaming

[Bison]

More people need to watch the Terminator series and a Space Odyssey. 

[The Devlish Redhead]

Do we have androids?

 

I know there's a robot in Japan that they have used for acting. She just sits and talks. But she's not autonomous

 

Ah found her

 

http://thecreatorspr...-stage-in-japan

[mybudgee]

it will be quite terrifying, when the day comes that we cannot tell the androids from the people. thankfully it isn't today.

I hope I'm long dead

[Decepticon Leader Sully]

Do we want to take the chance

[Voxr]

it will be quite terrifying, when the day comes that we cannot tell the androids from the people. thankfully it isn't today.

I hope I'm long dead

What's wrong with androids?

[Decepticon Leader Sully]

What's wrong with androids?

 

Yeah?

[Dermain]

Sure but earthworms do not have higher cognitive processes, we do. So outsmarting humanity might still be a bit demanding. And even it would and had access to internet there is still the matter of producing things it wants. The world is still too primitive for AI to rule it.

 

Personally I think a AI could help humanity to ascend. With this pace we are just killing ourselves and the planet.

 

The easiest way to outsmart humanity is with fear. 

 

i made a thread about AI with this title a few months ago

 

Didn't that get locked?

[mybudgee]

Didn't that get locked?

Nope. I linked it on page 3 I believe

[The Devlish Redhead]

i made a thread about AI with this title a few months ago

 

Got a link to that I'm interested

[mybudgee]

Got a link to that I'm interested

Promise not to revive it...

 

http://forum.bioware...oul/?p=17820537

[Decepticon Leader Sully]

[Decepticon Leader Sully]

any way taught i'd share this,

 

[mybudgee]

[Decepticon Leader Sully]

[The Devlish Redhead]

Promise not to revive it...

 

http://forum.bioware...oul/?p=17820537

 

NO I won't revive it but will read it.