Computers Are Incapable Of Creatively Writing Music

[Write4U]

Re: Tegmark

That's his Speculation. He has no actual Arguments to show what Emergence he is talking about

No, those are hard facts and he has actual arguments which support that hypothesis, and explained how any non-native influence on the native physics would be detectable and quantifiable.

It is you who is speculating about some mysterious "elan vital", which exists independent of "mind". And then you presume to call all other concepts as speculative, without offering a logic replacement of your speculations. C'mon man.....you appear hopelessly ...

 

[Vociferous]

I do find it funny how you blather on about irrelevancies, as if you think you're making a point worth listening to.

Yet you reply to every bit of it.

So Kasparov, Karpov, Fischer et al from the realm of Chess aren't in any way creative in the way they play their games? They haven't set the goal to win. Their brain is simply working out a method of doing some predetermined goal, right? Therefore no creativity? Just so I understand your point of view here: players of games are not creative, is that what you're saying?

And you're wrong about the "learning". AI such as AlphaGo Zero do learn. That is what they are very good at doing. Certainly not general learning but highly focussed. They were not told anything other than the rules to various games (e.g. chess, Go etc).

Chess players do not invent strategies wholesale. They learn from others and by trial and error (iteration) over time to improve their strategy. As stated, improvement and innovation are distinct from creativity. If they weren't, you fixing/souping up a car would be considered creative. Having a goal, like winning, isn't a new, valuable object. Generally, no, games do not require creativity, as I've defined it. If you have a better definition, we can compare citations.

AI does iterative innovation, akin to a chess player developing strategy. I'm fine with calling that learning, but then, is learning creativity? I don't think so. By definition, learning is acquiring preexisting objects, whether wholesale or by trial and error.

I'm sure you agree that creativity and innovation are not mutually exclusive. Innovation requires creativity, but just because one innovates does not mean they haven't also created what becomes the innovation. Even if the ultimate end goal is the same, it doesn't mean that all improvements are simply innovation in the absence of creativity. There is the creation of the new method, that hasn't been considered before.

Since only preexisting, valuable objects are innovated upon, all innovation requires some previous creativity, but they are distinct things. There's the new and valuable idea, and then there's innovating to implement that idea and bring it to fruition. So yes, innovation requires previous creativity, but since creativity doesn't require innovation, we know they are not the same thing. One necessarily proceeds the other. So saying they are not mutually exclusive doesn't really tell us much. Blurring the line between the two would seem to be equivocation, at best. An AI only recognizes something as superior because it has been programmed with that value judgement. Such value judgements do not originate in the AI. Methods are a means to an end, not an end product. A new method is just another way to accomplish the same ends, like a win. Just like in programming there are many methods to achieve the same goal, some more optimized than others. Even in US patent law, a method is only infringed upon if a single person (excluding devices) performs all the steps.

And when the better methods are themselves valuable objects? To other chess players, to itself, or Go players, the moves that AlphaGo (and subsequent versions) come up with are absolutely valuable. And AlphaGo Zero created them and then implemented them. Creation, resulting in innovation.
Now, one could argue that it is not AlphaGo Zero creating, but their programmers doing the creating through the tool of AlphaGo Zero... but since they didn't tell AlphaGo Zero anything other than the rules, and merely gave it the ability to play games against itself, for it to learn, and that it came up with strategies that no human player had yet considered as viable... I see that as a tough argument to make. But feel free. Could be interesting.

In the sense of creativity, it must be both new and valuable. "Better" is merely an improvement, not wholly new.
AI does not create anything wholly new and doesn't originate the parameters of a value judgement, nor even the goal. It just iterates faster than a human can. Like a car travels faster than a human can, that makes it a tool, not intelligence. The operative word in your description being "no human player had yet". That's just a matter of speed, not creativity.

If someone asks you to prove Fermat's last theorem (before anyone had actually proven it) then are you honestly saying that there would be zero creativity involved in finding the solution?

I also think we have different definitions of what it means to be innovative. I accepted your earlier definition of creativity including "valuable", but to me innovation also requires value judgement... judgement that it is better than what was already there, where creativity is in the arriving at it. And there would be value judgement in that as well.
I.e. Creativity is in coming up, without being told, with ways of getting from A to B - judged simply on whether they would get you from A to B. Innovation is in comparing the solution to what is already in play, and if judged to be superior then implementing it.

"Before anyone had" again implies speed rather than creativity. Seeing as the solution would be built upon incremental work of many people (including for the proof itself), we're talking about innovation. I've already said that innovation can be valuable, it's just not wholly new. Coming up with a path from A to B is something that can be achieved through brute force, without any creativity. You just compare every practically available path. There's judgement in the comparison, but you're not creating the paths. They already exist as a function of the relation between A and B. Recognizing a path is not creativity. Again, creativity also requires something wholly new. The solutions an AI arrives at already exist within the potential of the goal, a win within the rules. There's a finite number of ways to achieve that goal. It can just compare them faster.

Don't be so fatuous. Where did I say, or even suggest, that surprise alone makes something novel? It is what caused the surprise that you should focus on, and you know that. It is the move and the strategy that the AI came up with that stunned everyone, that was seen as creative and innovative.

Since surprising is a synonym of stunning, it seems you're objecting to the point more than the specific example. The point stands. Something that seems superficially novel, but already exists in the possible solution space, is not. "Was seen as" seems like an appeal to the people rather than reasoning.

The win is not necessarily the novel object, but I agree that it is certainly the impetus. The novel object in the case of AlphaGo Zero playing Go would be the strategy employed that noone had identified before, and in implementing it. It is also creating it because it never existed within its system at the outset. It knew the aim (to win) and the rules. That was it. So what is creativity if it is not the putting together of what is available in ways not done before, to arrive at a solution?

You're right. The win itself is not a novel object. The game and it's rules, including the parameters of a win, is the novel object.
"That no one had identified before" denotes speed and "had identified" denotes that it already existed, like within the possible solution space. You don't identify things that have yet to exist. Having never existed within the AI's system is just analogous to something it has yet to learn. Learning can be done by experience as well as direct acquisition. Learning, though, is not synonymous with intelligence. I can write an iterative program to find information it would take me much longer to learn on my own. The program itself does not learn. I learn from its output. With an AI, humans have to define the parameters of value and solutions that meet that criteria are fed back into the AI, as a basis for further improving solutions. Still all within the definition of innovation. Even if a game had never been won before, we'd still know what a win would require beforehand.

Sure, it is given the "win" as the target, just as improving a process might be the target at work. But that doesn't preclude the solutions arrived at being creative, being the result of creativity (e.g. the solution being something noone else had considered before). Sure, the recognising it as superior to the existing one at the time, and then implementing it, is the innovation. But it is more often than not a matter of creativity followed by innovation. Both. Not one or the other.
And given that noone has told the AI any pre-existing strategies, that it has come up with all the strategies it employs by itself, how is that not creating? It's not that it built on existing strategies by previous Go (or Chess) players fed into it in some brute-force approach that earlier computers may have done.

Again, "no one else had considered before" is not creativity. No one had considered the world was round before it was discovered, but discovering existing things is not creativity. Similarly, discovering a solution within the possible solution space is not creativity. Coming up with strategies in a game is a brute force process. The rules and the goal are sufficient for that. You don't need any preexisting strategies for a brute force approach. The rules and goal define the parameters of all available strategies.

Okay - so the idea that the infinite monkeys mashing typewriters wouldn't be creative even if they came up with the complete works of shakespeare. Got it. And I don't disagree.

Yes, and the complexity of an AI just obscures that. It's as if you were handed the complete works of Shakespeare and marveled at the creativity, but just never knew about the infinite monkeys mashing typewriters.

 

[Write4U]

Chess players do not invent strategies wholesale.

What is that supposed to mean, a reinvention of the game?
Well, check out the brilliantly creative moves by Magnus Carls this player

And in case this escaped your attention;

Which they can... just read up on Move 37 of Game 2 in AlphaGo's match with Lee Sedol, which stunned the world of Go.

AlphaGo came up with a move which had never been played in 2500 years of the oldest and most difficult game ever.

 

[Write4U]

Something that seems superficially novel, but already exists in the possible solution space, is not

You mean like making an illegal move that is not in the possible solution space? That would be creative, no?

Actually it is almost impossible to make an illegal move in Go.

How the Computer Beat the Go Master

With its breadth of 250 possible moves each turn (go is played on a 19 by 19 board compared to the much smaller eight by eight chess field) and a typical game depth of 150 moves, there are about 250150, or 10360 possible moves. This is a number beyond imagination and renders any thought of exhaustively evaluating all possible moves utterly and completely unrealistic.

Given this virtually illimitable complexity, go is, much more than chess, about recognizing patterns that arise when clutches of stones surround empty spaces.Players perceive, consciously or not, relationships among groups of stones and talk about such seemingly fuzzy concepts as “light” and “heavy” shapes of stones, and aji, meaning latent possibilities.

Such concepts, however, are much harder to capture algorithmically than the formal rules of the game. Accordingly, computer go programs struggled compared with their chess counterparts, and none had ever beat a professional human under regular tournament conditions. Such an event was prognosticated to be at least a decade away.

https://www.scientificamerican.com/article/how-the-computer-beat-the-go-master/

 

[Steve Klinko]

I'm sorry, but apparently you fail to understand that "bulk quantity" is a pattern.
Again, your making an incorrect assumption. I did not say wetness comes from water. I said wetness (water) is an emergent property of a collection of dry H2O molecules arranged in a specific pattern. Don't make false statements. That is not honest discussion.

What are you talking about. I said:
Your argument that "Consciousness Emerges from Neurons" is like "Wetness Emerges from Water" is also completely Incoherent. I can completely understand that when something is Wet or has Wetness, that it just has a bunch of Water on it. It makes no Logical sense then to say that when something has Consciousness it has a bunch of Neurons on it. What the heck are you talking about?

I clearly said "Wetness Emerges from Water" . I guess you are saying that "a collection of dry H2O molecules arranged in a specific pattern" is not Water. Your Incoherency continues to plague you.

 

[Sarkus]

If you reject my definition then there is no argument anymore. The definition is the whole basis of the OP. Good, go in peace.

I do reject your definition, and you're not helping anyone understand how you arrive at it.

 

[Sarkus]

Yet you reply to every bit of it.

We each have our weaknesses.

Chess players do not invent strategies wholesale. They learn from others and by trial and error (iteration) over time to improve their strategy.

And the new strategy, in that it didn't previously exist, has thus been created.
I think the difference here is that you require of creativity that what is created be of an entirely different nature, whereas I simply require that it not have existed before. Thus innovation encapsulates a creativity focussed on that which already exists.

Generally, no, games do not require creativity, as I've defined it. If you have a better definition, we can compare citations.

Producing something that is new and useful would be a simple enough definition.
New: a strategy not previously considered.
Useful: it helps achieve a goal.
Voila. Creativity.

AI does iterative innovation, akin to a chess player developing strategy. I'm fine with calling that learning, but then, is learning creativity? I don't think so. By definition, learning is acquiring preexisting objects, whether wholesale or by trial and error.

Yet you claimed the AI isn't "learning". So are you now backtracking on that?
The AI in question does not have any previous strategies to use. It has the rules. It has the goal (to win). The rest it does itself. Given it had no strategy previously, and now it has a strategy, how is that not the AI being creative, in an admittedly focussed way.

Since only preexisting, valuable objects are innovated upon, all innovation requires some previous creativity, but they are distinct things. There's the new and valuable idea, and then there's innovating to implement that idea and bring it to fruition. So yes, innovation requires previous creativity, but since creativity doesn't require innovation, we know they are not the same thing. One necessarily proceeds the other. So saying they are not mutually exclusive doesn't really tell us much.

It tells us that it is possible to innovate methods of getting from A to B while also being creative about how one gets from A to B. Just because someone has set the goal, and just because there is an existing method, does not mean that every subsequent means of getting from A to B considered is merely innovation. That's what you seem to be missing.

Blurring the line between the two would seem to be equivocation, at best.

What nonsense of a red-herring.

An AI only recognizes something as superior because it has been programmed with that value judgement.

Just like my boss tells me what the goal is. And that precludes me from being creative about the solution? 'Cos that's the only value judgement applied to the AI in question: the goal. The rest is entirely up to the AI within the universe of possibilities in which it operates (i.e. the rules).

Such value judgements do not originate in the AI.

What value judgements are you talking about? The goal? Sure, that's given to the AI. How does that mean it's not able to be creative?
If you mean other value judgements, which ones are you referring to?

Methods are a means to an end, not an end product.

They can be, and often are, the end-product of creativity.

A new method is just another way to accomplish the same ends, like a win.

How does that stop the new method from not being one brought about through creativity rather than just innovation? If you only take the end product of the "win" as being where creativity can lie then you're missing out on pretty much everything creative that has ever happened.
You would dismiss Da Vinci, Galileo, et al all as being just innovators, right? Every sculptor, every artist, every musician, as simply innovators, right? Every novelist, every dancer, everyone who has probably ever been described as creative. Because they're all merely doing something in different ways than has been done before. They're just working off the shoulders of those who came before, or achieving a goal set by someone else.

In the sense of creativity, it must be both new and valuable. "Better" is merely an improvement, not wholly new.

How can there be improvement if there is nothing there to begin with. With the AI in question, there is no pre-programmed strategy. There are the rules (the environment), and the goal.

AI does not create anything wholly new and doesn't originate the parameters of a value judgement, nor even the goal

Yes, it does originate something new: a strategy not previously seen before.

It just iterates faster than a human can.

So what. It got there first. Does that mean it can't be creative?? It created what noone previously had.

Like a car travels faster than a human can, that makes it a tool, not intelligence. The operative word in your description being "no human player had yet". That's just a matter of speed, not creativity.

Yet if a human had gotten there first you would almost certainly have attributed it to creativity! Although perhaps not, given that you don't think any players of games have been creative in their endeavours, and I await your answer with regard pretty much all artistic endeavour to date.

"Before anyone had" again implies speed rather than creativity.

Garbage. The two are mutually exclusive. One can be fast and creative. If I create something before anyone else had, does that preclude me from having been creative? No. So stop talking garbage.

Seeing as the solution would be built upon incremental work of many people (including for the proof itself), we're talking about innovation.

You're once again making creativity and innovation mutually exclusive. Where you see purely innovation, I see someone looking at something in a way noone else has (creating), putting together a method, then developing it. Not just innovation, and not just creation. But both. Sure, if someone had gotten 95% of the way, and all it needed was a slight tweak, then one could say that the person who put in the most effort, and first saw the possibility of the method, was the creative one and subsequent people merely innovating it.

I've already said that innovation can be valuable, it's just not wholly new.

It can be, even if the end goal is the same! Just because your goal is to get from A to B, and you consider improvements in method as innovation, for example, doesn't preclude new methods from being arrived at creatively. Just because in chess, or Go, the rule set/universe is limiting does not mean that all endeavours within it are therefore mere innovation. Again, that is saying that all endeavours within the wider ruleset of our own universe are equally just innovation.

Coming up with a path from A to B is something that can be achieved through brute force, without any creativity. You just compare every practically available path. There's judgement in the comparison, but you're not creating the paths. They already exist as a function of the relation between A and B. Recognizing a path is not creativity.

Everything anyone will ever do could be simulated by brute force, given a simulator of sufficient power and size. Is the simulator creative? If not, then how would whatever people do ever be considered creative, given that the simulator already recognised it as possible?
Whether something is done through brute force or efficient elegance, whether through mechanical or biological means, isn't of importance. If a brute-force machine can create strategies that it never previously had, then how is it not being creative?
Sure, not exactly creative on the scale that a person could be. Not as efficient either in power of speed as a person could be. But they are still creating something new (a strategy) that they didn't have before, that is of value to them - even if that value judgement (the win) was given to them to use. All you are doing is dismissing the fundamental core of what it is to be creative by requiring it to be a display of complex creativity. There's likely orders of magnitude difference on the scale, but it is nonetheless creativity.

Again, creativity also requires something wholly new. The solutions an AI arrives at already exist within the potential of the goal, a win within the rules. There's a finite number of ways to achieve that goal. It can just compare them faster.

Sure. And one could equally say that everything... and I do mean everything ... a human does already exists within the potential of the universe, the only difference in that we set out own goals (or at least most of us do). Are you perhaps suggesting that creativity is in the setting of those goals? If not, then how is goal-setting relevant, and if not then how is anything anyone ever does, human or otherwise, ever creative given that, as you seem to argue, the solutions we arrive at already exist within the potential of the universe?

The point stands. Something that seems superficially novel, but already exists in the possible solution space, is not. "Was seen as" seems like an appeal to the people rather than reasoning.

Nothing that ever happens is creative. I get it. Everything that happens is, by definition, already existent in the possible solution space. Thus nothing creative. Ever.
Thanks.

I'll stop there, lest I repeat myself more than I already have, and while you remain as unpersuasive as you are.[/quote]

 

[Steve Klinko]

I suppose that the biggest hang-up people have with this OP is that they don't understand a Conscious Experience or they reject the concept of a Conscious Experience. The key is in the Conscious Experience of the Music. Creativity can only come from a Conscious Mind, and Creativity never comes from Algorithms and Calculations unless you realize that the Creativity is possibly in the Minds of the Programmers. But the Music is still just Algorithms and Calculations. The Creativity is in the Programming and not in the Music.

 

[Write4U]

What are you talking about. I said:
Your argument that "Consciousness Emerges from Neurons" is like "Wetness Emerges from Water" is also completely Incoherent. I can completely understand that when something is Wet or has Wetness, that it just has a bunch of Water on it. It makes no Logical sense then to say that when something has Consciousness it has a bunch of Neurons on it. What the heck are you talking about?

But I never said any thing like it. All these analogies are completely false.

I clearly said "Wetness Emerges from Water" . I guess you are saying that "a collection of dry H2O molecules arranged in a specific pattern" is not Water. Your Incoherency continues to plague you.

Your "slippery of thought" is a result of convoluted brain activity. Your neurons are doing a jig .
Must be an emergent phenomenon of your brain neurons organized in a specific pattern.......

 

[Steve Klinko]

But I never said any thing like it. All these analogies are completely false. Your "slippery of thought" is a result of convoluted brain activity. Your neurons are doing a jig .
Must be an emergent phenomenon of your brain neurons organized in a specific pattern.......

Thank You

 

[billvon]

The Creativity is in the Programming and not in the Music.

Nope. Programmers do not program neural networks. They set up the initial network, then they train them. What comes out of them is not what the programmer intended - it's what the network was trained to do.

A lot like people.

 

[Steve Klinko]

Nope. Programmers do not program neural networks. They set up the initial network, then they train them. What comes out of them is not what the programmer intended - it's what the network was trained to do.

A lot like people.

Neural Nets are Configured using Algorithms that Programmers have developed and have Programmed into the Neural Nets.

 

[billvon]

Neural Nets are Configured using Algorithms that Programmers have developed and have Programmed into the Neural Nets.

Nope. The neural net is configured through learning.

Take your brain for example. It is initially grown via genetic instructions, just as a neural network is created using basic silicon structures and programming that configures that structure.

Once you are born you start to learn. Your brain is now more than the genetic instructions. Likewise, once a neural network begins training, it is now more than its physical structure.

 

[Steve Klinko]

Nope. The neural net is configured through learning.

Take your brain for example. It is initially grown via genetic instructions, just as a neural network is created using basic silicon structures and programming that configures that structure.

Once you are born you start to learn. Your brain is now more than the genetic instructions. Likewise, once a neural network begins training, it is now more than its physical structure.

A Neural Net is not a Brain. A Neural Net is just a Computational Tool. Neural Nets are Configured, only Brains Learn. We only say Neural Nets learn because of Convenience and Convention.

 

[Write4U]

A Neural Net is not a Brain. A Neural Net is just a Computational Tool. Neural Nets are Configured, only Brains Learn. We only say Neural Nets learn because of Convenience and Convention.

Let's try to find some common ground.
Is a Brain a Neural Net(work)? If you disagree with that, you are trying to rewrite the entire scientific definition of what qualifies as neural network . There are biological neural networks and the there are artificial neural network, but by definition they are both neural networks, ok?

Neural circuit

A neural circuit is a population of neurons interconnected by synapses to carry out a specific function when activated.[1] Neural circuits interconnect to one another to form large scale brain networks.[2] Biological neural networks have inspired the design of artificial neural networks, but artificial neural networks are usually not strict copies of their biological counterparts.

https://en.wikipedia.org/wiki/Neural_circuit

Can we agree that both human brains and computers employ forms of what we loosely call neural networks? I am trying to present "hard facts", not "hard questions".

If not then "never the twain shall meet" and all discussion of "common denominators" in AI and all other forms of "biological intelligence" becomes moot.

 

[Sarkus]

A Neural Net is not a Brain. A Neural Net is just a Computational Tool. Neural Nets are Configured, only Brains Learn. We only say Neural Nets learn because of Convenience and Convention.

No, we say NNs learn because that is what they do, and they mimick the way it is thought that our brains learn. It is convenience and convention because, when you come across something, it is both convenient and convention to call it what it is.

How do you think humans learn? What is it that defines the "learning", in your view?

Take the following example:
Person A doesn't know or recognise what a Quyon is. Person B does. B shows A a series of cards, some containing a Quyon (among other things), some not. Person A merely has to say whether they think a Quyon is somewhere on the card being shown them, and person B will say whether they are correct or not.

Gradually, through a process of trial and error, and reasoning, Person A can gradually start to identify what a Quyon looks like on the cards, right? E.g. the "i was told it was on this card, but not on this one, therefore it could be anything that is on both..." etc.

Would you accept that if person A can identify correctly, say 99% of the time, whether a card has a Quyon on it or not that they have learnt what a drawing of a Quyon looks like on a card?

 

[Steve Klinko]

Let's try to find some common ground.
Is a Brain a Neural Net(work)? If you disagree with that, you are trying to rewrite the entire scientific definition of what qualifies as neural network . There are biological neural networks and the there are artificial neural network, but by definition they are both neural networks, ok?

Neural circuit
https://en.wikipedia.org/wiki/Neural_circuit

Can we agree that both human brains and computers employ forms of what we loosely call neural networks? I am trying to present "hard facts", not "hard questions".

If not then "never the twain shall meet" and all discussion of "common denominators" in AI and all other forms of "biological intelligence" becomes moot.

In common usage when we say Neural Network, we are talking about Artificial Neural Networks, and you know that. But this is the type of thing that you do. Of course Brains have Neural Networks, not to be confused with ANNs. But ANNs are not Brain Neural Networks. Brain NN do not work like ANNs. You are fooling yourself if you think they do. ANNs are always only a first order (maybe Zeroth order) simulation of Brain NN.

 

[Write4U]

Would you accept that if person A can identify correctly, say 99% of the time, whether a card has a Quyon on it or not that they have learnt what a drawing of a Quyon looks like on a card?

Not just humans , but lemurs can learn some very abstract mathematical concepts as can artificial intelligences. I showed you how a GPT3 designed avocado shapes armchairs, mere from the command; "design an avocado shaped chair".
Did you see the video of the creative ways it design a whole series of avocado shaped chairs?

 

[billvon]

A Neural Net is not a Brain. A Neural Net is just a Computational Tool.

So is a brain. A brain is a natural neural network.

Neural Nets are Configured, only Brains Learn.

From the Wiki entry on neural networks:

"The utility of artificial neural network models lies in the fact that they can be used to infer a function from observations and also to use it. Unsupervised neural networks can also be used to learn representations of the input that capture the salient characteristics of the input distribution, e.g., see the Boltzmann machine (1983), and more recently, deep learning algorithms, which can implicitly learn the distribution function of the observed data. Learning in neural networks is particularly useful in applications where the complexity of the data or task makes the design of such functions by hand impractical."

 

[Write4U]

In common usage when we say Neural Network, we are talking about Artificial Neural Networks, and you know that. But this is the type of thing that you do. Of course Brains have Neural Networks, not to be confused with ANNs. But ANNs are not Brain Neural Networks. Brain NN do not work like ANNs. You are fooling yourself if you think they do. ANNs are always only a first order (maybe Zeroth order) simulation of Brain NN.

Again you just make unfounded assumptions. Nobody thinks that they are the same. It is ludicrous to even mention that as some kind of subtle distinction.
In fact it is the reason why the two systems have different prefixes. A metal car is not a wooden cart, but they are both transportation vehicles.

But both ANN and BNN are Neural networks, no? Else why would both have the term "neural network" in their name? I think you are missing the point that they are not talking about what separates them but what they have in common, i.e. data processing via a networked organizational patterns, instead of brute force.

Again, ask a GPT3 if it is conscious. If it answers in the affirmative, are you going to tell it is lying? It'll tell you; "Why would I lie to you", and look at you with incredulity.....

p.s. you are wrong also in the assumption that there no neural networks which simulate human brain networks.
If you are still at that level of understanding, you are decades behind the times.