18 réponses à ce sujet

[Obadiah]

A computer program convinced more than 33% of testers that it was a 13 year old kid, who was possibly speaking a second language.

A computer just passed the Turing Test in landmark trial
http://www.washingto...ial/?tid=pm_pop

[EDIT]
Ugh... wrong forum.

[Farangbaa]

[EDIT]
Ugh... wrong forum.

 

Glad this happened

[KaiserShep]

Throw it out the airlock.

[DeinonSlayer]

So any time you asked it something it couldn't figure out the default response was to blame the language barrier?

[Kabooooom]

I am underwhelmed by this and the backlash from the AI scientific community seems to be in agreement with me. This was not that impressive at all. We aren't even close.

Edit: originally I criticized the Turing test and compared it to the anthropomorphic hogwash that is the mirror test, but it seems that Turing himself did not even propose that his test could truly show intelligence, and overeager AI researchers have taken it to a conclusion that is actually criticized by the field in general. So, props to Turing, shame on people who design sh*tty methodology and draw inappropriate conclusions from them.

Let me know when an AI asks "does this unit have a soul?"

[Jukaga]

I hope mechanical AI is impossible, and it likely is. Super-processing ability does not equal self awareness and self-determination. The most we should even strive for would be idiot-savant expert systems with pseudo-personality overlays for interaction. Why anyone would want to actually build an artificial mind is beyond me.

[KaiserShep]

I'd like to have an AI of my own. I'd basically just have it watch movies. I'd start with the classics, like Short Circuit, Toy Story, Terminator, Terminator 2, The Day the Earth Stood Still (the good one not the shitty Keanu one), etc..

[sH0tgUn jUliA]

The test proved how stupid some people are.

[Reorte]

I hope mechanical AI is impossible, and it likely is. Super-processing ability does not equal self awareness and self-determination. The most we should even strive for would be idiot-savant expert systems with pseudo-personality overlays for interaction. Why anyone would want to actually build an artificial mind is beyond me.

Anything that can occur naturally can be created artificially, so I'm about as certain as it's possible to be that it isn't impossible. We're nowhere near doing it though. I agree that there's probably not much point in building one but I don't think it's as straightforward as that - you say "super-processing ability does not equal self awareness and self-determination" but I'm not convinced that it isn't possibly an outgrowth of sufficiently complex super-processing ability. The creation of an AI by accident like the geth doesn't sound completely implausible.

[Kaicel]

Let me know when they've passed the Voight-Kampff test. 

[Farangbaa]

Why anyone would want to actually build an artificial mind is beyond me.

 

To see if the Catalyst is right, of course.

[dreamgazer]

Why anyone would want to actually build an artificial mind is beyond me.

 

Never underestimate either the ego or the ambition of the organic mind.

[KaiserShep]

If it gets out of hand, we can simply block out the sun and kill its source of power.

[MassivelyEffective0730]

If it gets out of hand, we can simply block out the sun and kill its source of power.

 

Unless of course, it gets its power from the interior of the Earth. Then you're kinda boned.

[Kabooooom]

I hope mechanical AI is impossible, and it likely is. Super-processing ability does not equal self awareness and self-determination. The most we should even strive for would be idiot-savant expert systems with pseudo-personality overlays for interaction. Why anyone would want to actually build an artificial mind is beyond me.

I'm not sure if you were actually proposing that artificial intelligence acheiving sentience and sapience is impossible, or if you just think we shouldnt try to build one for ethical reasons. So, this post isn't necessarily directed in response to you, but it is just a general response to the thought that AI consciousness is impossible- which is a sentiment that I have seen echoed a surprising amount on these boards.

Although we dont have a true theory of consciousness, we know a hell of a lot about neurology. We know enough that we can actually predict that it shouldn't matter - there should be no difference between an organic system that processes information sufficiently and with enough complexity that awareness is present - or a synthetic one. The brain, presumably, is purely mechanistic. There is no ghost in the machine. To propose that only an organic mind can be sentient or sapient is, well, I'm not sure how you would describe it - biocentrist? It's logically flawed. The concept of some sort of "uniqueness" to the organic mind reeks to me of the same sort of BS that vitalism did. I'm convinced that people who staunchly believe that machine intelligence is impossible just don't know enough about neurology to realize why the method of processing the information shouldn't matter at all.

So yes, I'm quite certain a true AI can be created. But we aren't even close to that yet. We literally cannot even model the complexity of a Drosophila brain via artificial neurons. I've never understood why AI researchers attempt to grasp farther than their reach. Start small, model a simple neural network in entirety and THEN attempt an AI.

The problem, I think, is that until relatively recently AI researchers and neurologists didn't really communicate across disciplines. Hence, for nearly forty years AI research went in the complete wrong direction. But, in the past five years or so I've seen several cross-disciplinary studies that seem to be a step in the right direction. I hold out hope that we will see the creation of a true AI within my lifetime.

The question, though, is should we create one? The entire process is hugely ethically dubious. But, it doesn't matter, because we can and we will. We're humans, and we don't stop ourselves from doing stuff like that. We will do it, and we will deal with the consequences afterward...probably pretty poorly, for awhile.

[Farangbaa]

Of course we should, we'd create life out of nothing.

 

Only God has done so before.

[Eleinehmm]

Of course we should, we'd create life out of nothing.

 

Only God has done so before.

 

Considering that most people I know work in the weak AI framework, we are not even trying

[Eleinehmm]

So yes, I'm quite certain a true AI can be created. But we aren't even close to that yet. We literally cannot even model the complexity of a Drosophila brain via artificial neurons. I've never understood why AI researchers attempt to grasp farther than their reach. Start small, model a simple neural network in entirety and THEN attempt an AI.

The problem, I think, is that until relatively recently AI researchers and neurologists didn't really communicate across disciplines. Hence, for nearly forty years AI research went in the complete wrong direction. But, in the past five years or so I've seen several cross-disciplinary studies that seem to be a step in the right direction. I hold out hope that we will see the creation of a true AI within my lifetime.

The question, though, is should we create one? The entire process is hugely ethically dubious. But, it doesn't matter, because we can and we will. We're humans, and we don't stop ourselves from doing stuff like that. We will do it, and we will deal with the consequences afterward...probably pretty poorly, for awhile.

 

 

Do you work in a AI subfield per chance ? I contest the fact that we can.