AI Concerns

[Paradoxum]

I already said I don't know what a Humanist is, so I don't know.

Fair enough.

I believe in that too. So we never need to worry about AI, only worry about the humans who program AI, because there can never be actual AI.

AI can go beyond the creators.

 

[Chesterton]

AI can go beyond the creators.

How can you get consciousness from a flow chart?

 

[2PhiloVoid]

Where does it say that?

Revelation 13:14-15...but don't read it in English. Read it in the Greek, Para.

I'm okay for now.

Well, I'm glad to hear that, financially speaking of course.

 

[Jack of Spades]

How can you get consciousness from a flow chart?

You don't need an actual consciousness, just something that behaves like consciousness. Isn't the fact that we can build a machine which plays chess better than any human already a proof enough that we can create something which is either smart, or acts in a way that it's better than "smart".

Just imagine people in say, 1500's trying to build a machine which beats them in a chess.... It would have been totally hopeless and now we can do it.

 

[Chesterton]

You don't need an actual consciousness, just something that behaves like consciousness. Isn't the fact that we can build a machine which plays chess better than any human already a proof enough that we can create something which is either smart, or acts in a way that it's better than "smart".

Just imagine people in say, 1500's trying to build a machine which beats them in a chess.... It would have been totally hopeless and now we can do it.

Something that behaves like a consciousness? What behaves like a consciousness? Nothing. Consciousness is unique in the universe. I'll be impressed only when we can build a machine which is able to say "___ you, I don't feel like playing chess!" Or which says it wants to play chess, and then changes its mind.

When a machine can be as stupid, emotional, irrational and fickle as a human, without having been programmed to decide to be, then I'll believe in AI. But then that's a contradiction in terms.

 

[keith99]

You don't need an actual consciousness, just something that behaves like consciousness. Isn't the fact that we can build a machine which plays chess better than any human already a proof enough that we can create something which is either smart, or acts in a way that it's better than "smart".

Just imagine people in say, 1500's trying to build a machine which beats them in a chess.... It would have been totally hopeless and now we can do it.

Depending on what one considers beating humans in chess it may well have been possible in 1500! Is it beating the man on the street? An average Chess Player? A good player? Or is it beating the reigning human world champion?

The first may have been possible, though difficult, even that far back. The problem is that to be impressive enough to be worth a try even back then fell somewhere between beating an average and beating a good player and that was very doubtful.

 

[Jack of Spades]

Depending on what one considers beating humans in chess it may well have been possible in 1500! Is it beating the man on the street? An average Chess Player? A good player? Or is it beating the reigning human world champion?

The first may have been possible, though difficult, even that far back. The problem is that to be impressive enough to be worth a try even back then fell somewhere between beating an average and beating a good player and that was very doubtful.

Quoting myself: "machine which plays chess better than any human" - meaning beating the best of humans.

But, yeah. If we start stretching the definitions (and ignore the timeline when chess was first played) in Hellenistic era humans could have theoretically speaking built something like a waterwheel-propelled hammer which strikes randomly the chessboard, and repeat the game enough times vs a total idiot who can only move the pieces randomly and the "robot" would probably win one in a 8 trillion games when the pieces happen to jump in correct places enough times. I guess by some stretched definition that could have been called "being capable of playing chess"...

 

[Jack of Spades]

Something that behaves like a consciousness? What behaves like a consciousness? Nothing. Consciousness is unique in the universe. I'll be impressed only when we can build a machine which is able to say "___ you, I don't feel like playing chess!" Or which says it wants to play chess, and then changes its mind.

When a machine can be as stupid, emotional, irrational and fickle as a human, without having been programmed to decide to be, then I'll believe in AI. But then that's a contradiction in terms.

You're not asking very much to be impressed. Even I could program a machine like that, it would only need a couple of well hidden variables* which decide when it wants to play and when to change it's mind, so it would appear to behave spontaneously.

On a surface level, that would appear very human-like randomness and irrationalism. And that's not even an AI, it's just very basic usage of variables.


* = For example, a variable which counts microseconds in a clock and compare those to humidity of air, and with certain combinations, decide against or for it. Or simply a random generator.

 

[Chesterton]

You're not asking very much to be impressed. Even I could program a machine like that, it would only need a couple of well hidden variables* which decide when it wants to play and when to change it's mind, so it would appear to behave spontaneously.

On a surface level, that would appear very human-like randomness and irrationalism. And that's not even an AI, it's just very basic usage of variables.


* = For example, a variable which counts microseconds in a clock and compare those to humidity of air, and with certain combinations, decide against or for it. Or simply a random generator.

So you could program it to appear to behave spontaneously.

I'm not sure you know what the words "program", "appear" and "spontaneous" mean. You've made my point better than I could.

 

[Jack of Spades]

So you could program it to appear to behave spontaneously.

I'm not sure you know what the words "program", "appear" and "spontaneous" mean. You've made my point better than I could.

I just told you how to build machine which behaves as you wanted a machine to behave in order to be impressed. It's possible, and not even very difficult. It's not conscious or spontaneous, but it behaves like it is. Just with another kind of mechanism.

It's the same with an AI, it doesn't have to be conscious or intelligent in the very same sense as human is, it only needs to achieve the same things as consciousness and intelligence can. And it already can in many ways.

Feel free to deny what is already possible if you feel like it. Not my loss.

 

[keith99]

Quoting myself: "machine which plays chess better than any human" - meaning beating the best of humans.

But, yeah. If we start stretching the definitions (and ignore the timeline when chess was first played) in Hellenistic era humans could have theoretically speaking built something like a waterwheel-propelled hammer which strikes randomly the chessboard, and repeat the game enough times vs a total idiot who can only move the pieces randomly and the "robot" would probably win one in a 8 trillion games when the pieces happen to jump in correct places enough times. I guess by some stretched definition that could have been called "being capable of playing chess"...

Your knowledge of what the ancient Greeks could do is sorely lacking. My personal favorite is that they proved the Earth is a sphere, not just that they accurately calculated the diameter.

 

[Jack of Spades]

Your knowledge of what the ancient Greeks could do is sorely lacking. My personal favorite is that they proved the Earth is a sphere, not just that they accurately calculated the diameter.

I know plenty. People in antique could do and think many things which had to be re-invented centuries later because they had been lost. I was being tongue in a cheek.

 

[keith99]

I know plenty. People in antique could do and think many things which had to be re-invented centuries later because they had been lost. I was being tongue in a cheek.

Ah, POEs law got me this time.

 

[dysert]

I think the idea is that the Super AI can make more advanced copies of itself, or it can change it's own code. It doesn't need human fallibility.

This is the stuff of science fiction.

 

[Noxot]

science fiction often turns into reality

 

[Jack of Spades]

This is the stuff of science fiction.

Spoiler: So was this not long ago

 

[Jack of Spades]

Ah, POEs law got me this time.

I was actually left thinking that building a mechanical chess machine which could go for classic 3 move Scholar's mate - opening (gets beginners plenty of times) wouldn't be that difficult. Complicated yeah, but it would be possible pretty early in technological development, if the board was specifically designed for the machine.

That would hardly be an AI, rather an automation, but anyways it would be a non-human chess player, capable of beating the worst of human players who have just learnt the rules.

 

[Noxot]

who are we kidding? what is the human but a complex and limited amount of variables. that is the nature of being finite. thus the more capacity one has the more "intellect" it could have. think of a more basic kind of lifeform. it eats things through one end and poos it out the other and in this process it gathers what it needs to keep on going. in some ways humans are just highly complex versions of the most basic things of reality.

of course we spiritual people believe that union with God is our goal and find that the vanity of existing just to exist has no eternal worth due to it being limited. but if everything is a seed or a game that grows up or plays endlessly to God and God to us then maybe a super AI is the very thing that is gonna set humans straight, because they can't seem to play nice with each other yet. so if an super AI ends up being like a good mother then I would say that we could do worse.

what if super AIs are the ones who are gonna make all kinds of things possible for us that we would otherwise never have enough time to come up with before we are destroyed by natural causes such as our sun exploding or a comet hitting the earth? to shoot ourselves in the foot by limiting super AIs is imo wrong, it could result in the very thing we desired to prevent! so there is always going to be risk in reality so why not have a bit more hope rather than fear? fear is never going to make earth into a utopia.

being concerned for our continued existence is not really gonna help if indeed there are too many variables and possibilities that can occur.

 

[Jack of Spades]

who are we kidding? what is the human but a complex and limited amount of variables. that is the nature of being finite. thus the more capacity one has the more "intellect" it could have. think of a more basic kind of lifeform. it eats things through one end and poos it out the other and in this process it gathers what it needs to keep on going. in some ways humans are just highly complex versions of the most basic things of reality.

of course we spiritual people believe that union with God is our goal and find that the vanity of existing just to exist has no eternal worth due to it being limited. but if everything is a seed or a game that grows up or plays endlessly to God and God to us then maybe a super AI is the very thing that is gonna set humans straight, because they can't seem to play nice with each other yet. so if an super AI ends up being like a good mother then I would say that we could do worse.

Yes, more or less so, our biological self is little more than an extremely complex machine. In my opinion, we can believe in the soul and spirit while still recognizing this. The fact that we're biological computers, doesn't mean that's all there is to us in spiritual sense. That's a different question.

what if super AIs are the ones who are gonna make all kinds of things possible for us that we would otherwise never have enough time to come up with before we are destroyed by natural causes such as our sun exploding or a comet hitting the earth? to shoot ourselves in the foot by limiting super AIs is imo wrong, it could result in the very thing we desired to prevent! so there is always going to be risk in reality so why not have a bit more hope rather than fear? fear is never going to make earth into a utopia.

One thing that's good with having an AI run things is, it doesn't have the vices of humanity, it's objective. Having an AI drive a school bus might sound scary, but there are many humane bad things an AI doesn't do. It doesn't get drunk, it doesn't cowardly abandon a tough situation to save it's own skin, it doesn't panic, it doesn't feel the pressure to hurry to end it's shift early and so on. Human judgement is clouded by things like that.

In all of it's amorality, an AI can end up being morally (even tho it doesn't recognize the concept) superior to many humans by just being objective. It doesn't have virtues, but it doesn't have vices either.

 

[keith99]

This is the stuff of science fiction.

You mean stuff like cell phones, cars that can drive themselves, medical devices that can see inside people and rockets that can take us to the moon?