I’m still not seeing any convincing argument that emotions are anything but orthogonal to sentience.

~J

*shrug* in real life, it's as faulty to assume that AIs will have emotions as it is to assume that they won't. no one's ever seen a sentient being that didn't (excepting humans with certain psychological problems), so you really can't make a claim either way.

in SR, there's a wider variety of intelligences to compare against; the world is chock-full of sentient beings besides humans and metahumans. all of them, thus far, have displayed emotion (though some of them, such as shadow spirits, display a different range of emotions from human normal). i think it's safe to assume that in SR, at least, AIs will probably display emotion.

regardless, if a program does display something that resembles emotion, my stance is that it's only mimicry if the program isn't self-aware.

untrue. alice haeffner died during what may have been the upload of her psyche. it's also possible that the AI that's taken her name merely took on aspects of Alice's personality; Alice is little more than a rumor, so it's hard to say exactly what she is and how she came to be.

I guess then you haven't read (any of) the novel(s) in which she was featured rather prominently? They made it pretty clear that she is in fact a ghost in the machine, retaining all her memories thought processes ect.

ah. no, i haven't. shadowboxer lit on fire my ability to read SR novels. she did die during the process of becoming an AI, though, right? which leaves the possibility open that her AI is more than the sum of the data that existed in her brain.

Oh da joy, geeks at their best. Sometimes I forget why I started shadowrun,and the awnsers to this thread remind me. Because somethings are never right,somethings are never wrong and sometimes you just have to make do with what you have.

Alice isn't supernatural, she's entirely technological at this point. A consciousness in the matrix.

...a consciousness that has no code. that doesn't seem very techological to me.

As stated earlier, its simply a collection of cohesive electrical impulses It started out grounded in a meat brain, now grounded in the matrix.

I'm not happy about it at all, in fact I wish It didn't exist because I think it's beyond cheezy and stupid, but this is what we're stuck with.

hmm, could it be that the sr matrix acts more like a electronic copy of the human brain then they want to admitt? could explain some stuff

What's really the question here is what the difference is between mimicing self awareness and being self aware? Aren't they really the same? I mean, an A.I may not really be self aware, but it thinks it's self aware...

Really, if you can't tell the difference, then there isn't one.

JaronK

*shrug* we're not really discussing practical differences. there really isn't any practical difference between a true, self-aware AI and a program that can mimic self-awareness to the point that no one can tell it's not. but if someone lies to you, and you never find out, they still lied.

The thing is, if someone lies by saying "I'm lying" then they're not really lying, now are they?

If a machine thinks it's self aware, it can't really be wrong, can it?

JaronK

in order for an intelligence to think it is self-aware, it has to be self-aware, yes. but try to prove that an AI actually thinks it's self-aware, and isn't just saying that because it's programmed to appear self-aware.

The thing is, how can you prove you're self aware? Maybe you're just saying you are. Maybe you just think you are because a lower level of your brain is telling you that you are, because it has evolved to do so. At some point, your intelligence lying to you enough that you're self aware is enough to make you so.

JaronK

the ability to question the presence of one's own self-awareness is not possible without self-awareness. i think, therefore i am. as i've acknowledged several times, it's difficult to prove self-awareness to others. a non-self-aware AI will not question its own self-awareness, or lack thereof, because it has no self to question. the act of questioning its self-awareness would be an indicator that it has made the leap from mimicker of self-awareness to an actually self-aware entity.

Many forms of AI have layers of programming. What if the higher level programming questions its own awareness, because a sub-routine was programed to do so? And how is that different from a human conciousness questioning its own awareness, when the question comes from your sub-conciousness? After all, what is the subconcious and instict, if not a series of low level sub routines?

JaronK

There's total speculation. We have no AI today, and the AI in shadowrun has never been addressed in such a fashion. Are you totally pulling that from your behind or is that relevant in some way?

You're addressing the difference between something that knows it's alive and something that is told it is. Big difference.

For the record, I'm a computer science major who's specialized in Artificial Intelligence. Depending on your definition of Artificial Intelligence, I've programmed one. It certainly wasn't self aware, but it was good enough to do a few things that surprised me, and I wrote more than 70% of the code (and oversaw the writing of the rest).

Now, there are a variety of possible structures and methods of creating AIs. With the one we went with, it had three layers. The first layer was an enourmous group of substrategies (this AI was designed to play a game), mostly that I programmed in, some that it created through formulae and recombination of existing substrategies. The second layer was an evaluation layer... it took each strategy, and determined which strategy was doing the best at any given time. It then chose the best strategy and sent it on to the third layer. The third layer took the guess sent by the second strategy, and used information from the first layer to figure out what the other players were doing, and modified the guess from the second layer so the bot wouldn't play something that it knew an opponent would counter.

Now, if you look over the whole code, you can figure where any individual bit of information comes from... exactly which line of code generated what. However, if you simply were the top layer of code (the third layer), to you what's happening is you're thinking up guesses, comparing them to what you think the other players will do, modifying your original guess, and going with it. As far as you know, you just magically thought up a number to work with.

What I'm saying is, if someone saw all your code, inside your brain, they could probably tell you where each individual thought you have comes from. But since you can't actually see the processes in your own brain from lower level areas of your mind (your instinct, your subconcious, etc.). Self awareness, then, is sort of a missnomer... it's the very fact that you're not aware of where all your thoughts come from that makes you think you're self aware. What is self awareness, after all, but mystery... the fact that you can't point to where in your own head that thought about how nice Davinci's painting was came from. Yet perhaps it came from a some instinctual part of your brain that was programmed to like colorful things because fruit is colorful and could be food, combined with a piece of your object recognition subconcious routines that liked the symmetry, because it was easier to process.

This is why AI can never be self aware according to it's coders. Yet it might think it's self aware... and shouldn't that be enough?

JaronK

Frued was a crackpot, and in the psychology world is pretty much universally acknowledged as such, though for some reason the art world still loves him.

With that said, no, I don't claim to have solved everything by myself... considering how long I've been studying this, I know I'm not the only one saying it. But whatever you want to call it (without going into spiritual collective crap), the fact is that the subconscious is by definition a group of mental impulses which your higher mind has little controll over or information about.

And seriously, what is instict but a series of instructions programed into your brain at birth? Isn't that the definition?

And isn't that what a series of subroutines is?

JaronK

so, because the definition of "self-aware" can be obfuscated, any program that mimics any emotion is now an AI? that seems like a pretty flimsy definition--at least as useless for externally diffing AI and not-AI as my own standards, and without the benefit of decent internal diffing.

Well, yes, I am dismissing most religeous bits out of hand, because there's no scientific proof. Instinct, however, is measureable and observable. Even if you go the religeous route, one could say that "God" is your programmer, but none the less... what makes you think you're self aware? Where is that coming from? If I could point to the exact part of your brain that makes you think it, you might say you're not really self aware, because you're programmed to think you're aware, right?

At least, that's essencially the arguement that people give for why A.I. can't be self aware... because somewhere, deep inside the programming, is the part that makes it think it's aware, and that's scripted, programmed, and thus somehow a lie.

But the truth is, self awareness is a contradiction. It is only the mystery, the lack of awareness of all of your self, that makes you think you're self aware.

At least, all this is the oppinion of one A.I. programmer. But having seen the building of A.I. in its infantile stages, I think self awareness is really only created by the complexity the mystery that comes from that complexity. This translates nicely in Shadowrun, where you create an extreamly complex learning agent (Semi-Autonomous Knowbot), then let it lose in the world for long enough for its complexity to obfuscate its code enough that even the original programmers don't know what it will do next. Thus, you have an A.I.

JaronK

The simple solution is to make a Knowbot that isn't programed to do anything. If it does something, it obviously has free will.

I'm not sure just "doing what you wouldn't expect" counts as free will. I know the last A.I. I programmed was for a game where you had to guess numbers while also guessing what numbers your 20 or so opponents might guess, to get the right number for that round. On a lark, I had it play another game where you chose either "Go" or "Not Go", where if you went and no more than 60% of the others went you'd win, but if you didn't go and more than 60% did, you won. While the agent didn't play as well in the second game, it did perform admirably enough, dispite never having been programmed to play that game at all (or even to guess anything other than numbers).

The way that particular agent learned, by the way, was by imitating others it saw in its world, then figuring out whether that particular immitation was a good idea or a bad one, which is very similar to the way a child learns. It was very successful.

JaronK

MFB: I just got done earning a 4.0 in Biopsychology, and I can tell you that some of what JK says is true. For some thoughts, we can tell precicely what area of the brain they originated in. I can tell you what neural pathways brought them there, and what types of processing occured at each point along the way.

If we break down the human brain far enough, the theory is that we can predict with great accuracy where each though originates. We're up against an ethical limit, not a practical one-- we're not allowed to vivisect working human brains to see how they work.

So, in theory, we could break down a human brain until we find the part that says: "I think, therefore I am."; and isolate it. By damaging or removing that section, we could remove a person's self-awareness, and then see if that's a core part of his or her sentience.

Actually strictly speaking of SR we do in fact have several true AI without auras.

true. AIs work on wildy different principles, though; it's probable that the basis for an AI's self-awareness is completely different from the basis for a human's self-awareness. heck, it could be layers.

Actually, the AI's basis for self awareness is the "x factor" that they don't understand in SR, even those who made it.

Mfb. You make a strong claim when you when you say that there is “concrete evidence that points to the existence of a soul-like entity that resides in all living beings.”

I have a signifignt amount of anecdotal evidence to suggest that this is true but not enough for me to believe it is true (I do not however believe it is false). I would like to know what evidence you have (ether threw a link, posted on this forum or a personal message as you se fit) I truly hope that you are correct.

At present the information I can acses shows the probabilities leaning towards sufficient examination of the human brain would allow us to identify the proses of eth subconscious mind as we would with the AIs mentioned by the likes of JaronK. This is more probable because I m aware of the existence of reproducible experiments that uncover the source of some types of thoughts and teat the rain hs not yet ben aped ell enuf to expect to find them all. Meanwhile I am aware of no reproducible test that shows the presence of a soul-like entity or any related phenomenon. I would love to be shown a counter example.

Edward

On a more SR note I don’t have any source books that describe an AI, dose it state that a AI has no aura or is this left unsaid.

Edward

it is, as far as i know, left unsaid. the "evidence" i was referring to applies only to SR, where all living creatures have an aura. in real life, i suppose you could say that stuff like Kirilian photography points to a part of living beings that we haven't yet identified, which could be construed as being non-conclusive evidence of the existence of souls. it could also be construed as a lot of other things, though; at this point, it's all guesswork.

One of the reasons I'm interested in A.I., in fact, is precisely because we can't take apart a human brain and watch what's going on when someone thinks, but we can watch what an A.I. is thinking while it operates. Thus, if we can create an A.I. that behaves exactly like a human, we can see what makes the human mind behave as it does. Perhaps even more interesting, if we can create a criminal A.I., we can find out what causes certain criminal or deviant behaviors.

One test, by the way, to determine of something is actually an A.I. is if you talked to it online, can you tell if it's a person or a computer? If you can't tell, it's an A.I.

JaronK

yeah. i mean, i love doing Matrix stuff, but it's only because i turn my brain off when i play.

Why in the world is this discussion still going on?

Let's consider VR2.0. An AI is at leastr comparable to an SK. An SK requires, at minimum, a Red 10 host, and the eqv. of a half-dozen or more programmers with Computer 12. (140-41).

"programming an SK is beyond the capabilities of player characters (Matrix, 147).

More fundamentally, "AIs cannot be created--they happen." (Matrix, 150).

The criteria is:

This thread isn't even two pages long yet

~J

Good for you.

95 responses for this is too much.

There are worthless threads, just like there are worthless questions (and people).

This is one of them.