Someone's claimed that they've created a SK., Not a true AI....yet. Options

[SpasticTeapot]

http://biz.yahoo.com/prnews/051202/clf017.html?.v=33

Cool, no?

[Drace]

Its basically a very advanced Know-bot/expert system. Some hospitals use something much akin to this, known as E.L.I.Z.A., which is a medical diagnostical bot.

The closest thing to a true AI is still only a hypothetical quantum computer that would use trinary, rather than binary (the AI from Stealth was based of this hypothetical computer), but even then, it would still only be semi-autonomous, for it would still need imput instructions, it wouldn't work on its own accord.

[Chibu]

Even if, eventually, you didn't need more input, it would still use alrogithms to determine the best course of action

The reason I (as a computer programmer) don't think that an AI is possible is that for it to be full an AI, it has to be able to make mistakes. Humans do use alot of information-based intupt to produce an output, but, with current computers, it's not really possible. What defines sentience, in my oh so humble opinion, is the ability to invent and to create beauty. A computer cannot create somethign new, it can, however, take what is already invented, and put it into the most efficient pattern to create something new.

To create an Intelligence, it would have to be able to do the following things (well, for me to recognize it as such anyway:

#1. Create Beauty: Any form is alright. Take programming. When asked "Why did you choose that piece of code?" I have answered, "Because it makes it beautiful. It's inventive. It makes the code LOOK good." it's not always "becuase it id the most efficient way to do it"

#2. Infer: To be able to come up with an answer withought a complete set of data. Such as, when reading a new word, I sometimes understand it becuase of the way it's used. Though, this applies to all situations

#3. Make Mistakes: Because sometimes it makes sense for two plus two to equal five.

#4. Emotion: I believe that emotion and sentience go hand in hand. If you can't be sorry for something, you are not sentient. If you can't owrry about anything, you're probibly not sentient.

#5. Self Aware: You have to know that you exist.

#6. Belief: You have to be able to believe in something that you don't know. For instance, I believe in a God that created everything, and I believe that my girlfriend is not cheating on me. Though, there is no proof of these, becuase they are things that are not possible to prove. Never the less, I still know them to be true.

I think that's all. I don't really know what I started talking about, but, that's not important. (Maybe I should add "Ability to ramble" to that list)

But, an SK is still pretty cool :P I'm glad that tech is progressing as fast as it is.

(EDIT: Wow, i just read that, and i'm not sure what train of thought led me to ramble about how AI can't exist. But, it pretty much proves that i'm sentient :P)

[Oracle]

According to your definition I am not an intelligence / alive. :(

[Arethusa]

I'm not even going to touch belief. Suffice to say I am agnostic, and apparently therefore not alive and or intelligent.

[Critias]

That wasn't me that said that. I'm not sure why your quote-thingie says Critias.

[Arethusa]

Crap, sorry about that. I'm not sure either.

[Critias]

I mean, it's no big deal, or anything. I wasn't offended, or feeling as though I was being illegally misrepresented, or whatnot. Just confused.

[Arethusa]

It was most likely a combination of extreme sleep deprivation, distraction, misreading Chibu, and then forgetting to correct it afterwards.

I guess this means you have to kill me now in accordance with forum law.

[Chibu]

Belief doean't have to be in a god, like i said in the example about the girlfriend.

and, yes, I've since realized that emotion is not necessarily a defining term. That is more of a list of "if someone has these things, then they are sentient. Otherwise, it's up to you to determine. And, also, with Belief, comes disbelief. So, in questioning me, you prove your sentience. :P

[Arethusa]

Yeah, but agnosticism isn't necessarily limited to beliefs about religion. One can take an agnostic stance on all of reality, at which point, it is quite broad enough to disqualify that criterion. Hell, if we've created an artificial intelligence that doubts its own existence and questions the nature of reality, I'm damn well ready to call it truly intelligent.

[Adarael]

By definition, agnosticism taken to an extreme constitutes a belief. Or even not to an extreme. Even if that agnosticism says, "There's a buncha crap out there that I'm never going to know about," it constitutes a general faith in the structures of your life and a willingness to act on the belief that they are consistent, *yet* by virtue of adaptability, not totally shut down when a pattern changes.

If you're so agnostic you care about nothing, that's just an apathy problem.

[Oracle]

[hobgoblin]

my main requirement for a AI is for the program to be able to add new code to itself to handle a new situation. to do that, it needs to be able to evaluate what is required of it in refrence to the actions attempted.

the rest is just social requirements for a "normal" person. a AI is by definition not a normal person...

if we ever manages to create a true AI it will be cold and calculating. and maybe even easy to confuse, like say how HAL was.

so let me take those points apart:

1. beauty is in the eye of the beholder. basicly its not a universial constant. beauty is more about reproduction then anything else. therefor a AI may well find a string of binary beautyful as it reprecent a set of perfect code :P

2. this can happen. what you need is a large/complex enough neutral pattern. an ability to run a simulation on top of said neural pattern where you basicly try to randomly fill in the blanks. and in the end the ability to save said simulations and then present them as theorys. the big thing is that it needs to be able to do the theorize, compare, evaluate loop without the need for human interaction. creating a theory is simple, comparing it towards the know facts is a diffrent story. evaluating how correct the theory is compared to the real deal is the hardest of all.

3. mistakes happen. allready today a computer can make misstakes (do what i wanted you to do, not what i told you to do kind of mistake). the trick isnt doing misstakes but learning that its not a useful way of doing what one is trying to do and rembering it. kids learn this way from a early age by putting shaped blocks into similary shaped holes.

4. emotions are just levels of hormones in our brain. a computer can nicely simulate that by increasing or decreasing some values based on diffrent input. so a computer can have emotions, but they dont have to be similar to what we see as emotions.

5. this is the big trick isnt it? what does it realy mean to be selfaware? to be able to look at one limb and say, this is me, look at another beings limb and say, this is him or her, we are two diffrent entitys? if so, this will be damned hard to do for a intelligence with no extremitys.

6. the moment an AI starts beliving in stuff, is the moment i turn it of. beliving without questioning is what have fueld the worst actions on this planet so far. belif can be a dangerus thing.

[nick012000]

[Critias]

Personally, I don't know why we weak, fleshy, humans, are still trying to make an AI. All it's going to do is find us obsolete, take over the electronics we use to survive in our day-to-day lives, and kill and/or enslave us.

If a bunch of pencil-neck programmer/scientist/geeks really really need to feel like a god by creating a new intelligent life form, they can hump 'till someone gets pregnant, like the rest of us poor schlups that want to steal a sliver of immortality by causing a thinking creature to come into being and call it "ours." There's no need to doom us all to extinction or a life of servitude to our robotic masters, over the whole thing, when "insert tab A into slot B, repeat" has been working for poeple forever and ever..

[Sicarius]

We believe in nothing Lebowski... NOTHING!

[Crisp]

[nick012000]

[Oracle]

There is an interesting analogy in there. Let us assume for a moment that we have been created by a powerfull entity whatever the name of that entity may be. Today a very large part of society does not believe in creationism anymore. And they do not believe in the existence of a creator.

Possibly at a day in the far future, long after the last human died in the robot wars, the AI-species created by human scientists will stop to believe in our existence. There has never been anything like a creator.

Has any sci fi author ever written about something like that?

I should add that I am not religious.

[hobgoblin]

[hyzmarca]

[hobgoblin]

true, it would be something like the worst case of split personality known :silly:

however, this would either require a brain "emulator" running on some machine, not very practical, or uploading the backup to a clone.

in either case, one starts to reevaluate the value of life...

[ShadowDragon8685]

[Kyuhan]