DeepMind's AlphaZero plays chess like a tornado in the junkyard

[In situ]

All the millions of chess moves already exist (in potentiality anyway), just as all the roads between N.Y. and Los Angeles exist.

Just because you are a Platonist does not mean everyone agrees with you.

 

[In situ]

All possible things exist in potentia just the same.

This is the same thing as saying reality is a chaos. There are only two way to achieve deising from chaos: impose rules, i.e. order, or wait an infinite long time. Since we do not have infinite long time available we need to impose rules on AlphaZero.

 

[In situ]

Do you think the OP is a theist? I don't know actually. I suppose he could be.

I am labeled as agnostic. Does it matter for you what I believe? If it does, I can p.m. you what my believes are. Just let me know.

 

[In situ]

I dont see how solving 2+3 is in any sense comparable to designing a thing or a method.

I do. But there is a difference in being designed to do something (a calculator) and being designed to design something to do (AlphaZero).

 

[Speedwell]

This is the same thing as saying reality is a chaos. There are only two way to achieve deising from chaos: impose rules, i.e. order, or wait an infinite long time. Since we do not have infinite long time available we need to impose rules on AlphaZero.

Right. I was just running with Chesterton's observation.

 

[In situ]

You design things. Saying it designed a "way" is a very loose use of the word design.

What makes you think only things can be designed? Are plans, i.e. ways, not designed?

 

[In situ]

I was just running with Chesterton's observation.

And I was just running with your observation. I didn't thought about it in that way before you pointed out the "obvious". I guess I should say thank you, that comment definitely helped to straighten out my own thinking.

 

[In situ]

you stated AZ didnt have a telos

Were did I state AlphaZero did not have telos?

 

[Chesterton]

The things the adversary can do in a dynamic fashion to the available routes of travel explode the problem beyond the capacity of pure brute force analysis.

You can complicate it all you want. You can add adversaries, road closures, bad weather, etc. It still won't mean the traveler created anything.

I dont see how solving 2+3 is in any sense comparable to designing a thing or a method.

I don't either. So why do you think AZ math work is? Just because it's a lot more complex?

 

[Chesterton]

What if you discovered a substantially similar algorithm working in nature?

Math is the universal language as they say. Pretty much everything can be described or explained with it. I would be very concerned if these AZ geeks created anything which didn't have an analog in nature. Then they might be doing magic.

 

[Chesterton]

So you claim they cannot be compared. Is it correctly understood your reason for this claim is that living beings are spiritual beings?

No, I just threw that in there as an added difficulty.

Just because you are a Platonist does not mean everyone agrees with you.

Before AZ was developed, it was already possible to make all the moves that are possible to make. Is that any better?

I am labeled as agnostic. Does it matter for you what I believe? If it does, I can p.m. you what my believes are. Just let me know.

Your belief personally doesn't matter to me, but yes, as I previously explained, I think the consideration of the origin of the rules of nature, of computing and of chess, is relevant to your argument.

What makes you think only things can be designed? Are plans, i.e. ways, not designed?

Yes you can say plans are designed. I can design a plan in my mind right now using just abstract thoughts. But you're comparing that (what AZ did) to biological evolution where tangible things are created, where the plans are made incarnate, so to speak.

 

[durangodawood]

You can complicate it all you want. You can add adversaries, road closures, bad weather, etc. It still won't mean the traveler created anything.

I don't either. So why do you think AZ math work is? Just because it's a lot more complex?

AZ is a machine for designing strategies for a difficult intellectual challenge. The machine created these strategies. They certainly didnt emerge from the minds of the human programmers.

 

[durangodawood]

Which is not a difference at all. AlhpaZero is given Monte Carlo Tree Search and the rules of chess, evolution is given genetics, i.e. natural laws.

To understrand what the tornado has to do with anything, i.e. is irrelevant here, see post #140.

You misread me.

Here's what I said; The real difference between AZ (or evolving life) and the tornado>747 is pre-existence of the algorithm (or DNA).

 

[durangodawood]

Dictionary definitions are in general not suitable in a philosophical discussion.

I'm not referencing the definition so much as the example they chose, which supports the idea of designing a system or method.

Do you think its improper to speak of designing a system or method?

 

[Chesterton]

AZ is a machine for designing strategies for a difficult intellectual challenge. The machine created these strategies. They certainly didnt emerge from the minds of the human programmers.

Okay, but that's just saying it's analogous to any machine man makes, like a bulldozer which can tear up a hillside, and the power involved in doing so didn't emerge from the hands of the men who designed the bulldozer.

 

[durangodawood]

Okay, but that's just saying it's analogous to any machine man makes, like a bulldozer which can tear up a hillside, and the power involved in doing so didn't emerge from the hands of the men who designed the bulldozer.

Sure, that makes sense.

But making a machine for creating intellectual achievements is really quite something, dont you think? It could open doors to amazing places.... or places we'd rather not go.

The bulldozer is basically a manpower-multiplier. More of the same. But a machine for intellectual feats could do things we cant even imagine... like the way AZ's style of play has taken chess experts totally by surprise.

And this is merely the absolute infancy of intellectual machinery.

 

[In situ]

This is one of the first post to this thread I have not really responded to until now. Mainly because my thinking about the topic had not been straighten out until now.

No, but if I have understood what you are saying you are comparing a piece of software with the theory of random selection.

Correct. Beside selection is not random unless in cases such as natural disasters. What makes you believe it is? I am asking since I have a hard to see how improved camouflage is a a random process. If selection is random it suggest camouflage does not work.

The software is designed to complete a task; in that sense it is volitional, i.e it has a task to accomplish and a preprogrammed ability to complete that task.

I agree, and I granted telos in the OP and stated it is irrelevant for the argument.

It does this by trial and error and by remembering and applying what it learns through this process. Evolutionary theory does not allow for a task to complete or a process of learning, or there being any sense of there being a predetermined goal (except in all cases by metaphor).

This is where you are incorrect. Evolution accounts for all of this. The task to complete is to survive in order to reproduce, i.e. this is the goal or purpose of Life. The learning is represented by selection, and the lessons learned is stored in the DNA of the offsprings.

What you are describing is a better argument for creationists than for those who believe in a blind process of non volitional random selection.

See post #140 for a refutation of this.

 

[In situ]

Okay, but that's just saying it's analogous to any machine man makes, like a bulldozer which can tear up a hillside, and the power involved in doing so didn't emerge from the hands of the men who designed the bulldozer.

There is a difference between a machine which do something and a machine which design something which in turn will do something. Let us call this 1st and 2nd order machines. The purpose of a 2nd order machine, like AlphaZero and evolution, is to create another machine, like a 1st order chess playing machine or new species.

Your bulldozer is a 1st order machine, controlled by humans who are 2nd order "machines". Human designed 1st order machine are built to extend our bodies capabilities beyond our own. Human designed 2nd order machines are not only built to extend machines capacities but also to extend our design capacity beoynd our own.

There is gross misunderstanding in this thread what the purpose, or given telos, of AlphaZero is. It seams to me people has conflate 1st order machine with 2nd order machines. All machines has telos, but AlphaZeros telos is not to play chess. Its telos is to design a 1st order chess playing machine which telos is, or become, to play chess. It is the telos of this 1st order machine which AlphaZero designed by itself - no human was involved in this, only randomness and selection.

Second order machines are machine which builds other machine with a telos which, in general, is not the same telos as the 2nd order machine itself has. If the telos is the same then the telos is know as self-replication. Which is the case of Life, i.e. the purpose of Life is to create more Life.

 

[klutedavid]

Googles AI engie DeepMine has been trained to play chess. They call the program AlphaZero. Only given the rules of chess it had to experiment by random trial and errors games with itself. After 4 hours traning it was matched against the best chess engine in the world; Stockfish. After 100 games, AlphaZero won 28 and lose zero time, with 72 draws

For those not initiated this is quite impressive. As quoted form Chess.com:

"This would be akin to a robot being given access to thousands of metal bits and parts, but no knowledge of a combustion engine, then it experiments numerous times with every combination possible until it builds a Ferrari."​

Which is similar to the junkyard tornado argument creationists like to use to "disprove"evolution. According to creationists claims, it is like DeepMind's AlphaZero learned to plays chess similar to a tornado in a junkyard would create a Boeing 747. I know some Creationist now will claim AlphaZero been made by an intelligent designer. I grant telos, but it misses the point; AlphaZero still had to figure it out by itself, by random trial and errors, how to play chess at a super human level (see post #136 and post #140).

I am curios what take creationist have on this; if randomness cannot create design, no matter what time is given, then what did cause AlphaZero to achieve a superhuman performance in chess in less than four hours time?

Errata: AlphaZero did not achieve super human performance in four hours, after four hours training it was able to beat Stockfish. AlphaZero had to continue train fore a few more hours to reach super human performance.

N.B.
Post #81 - Clarification of my question.
Post #82 - Why AlphaZero is not like other chess engines
Post #86 #155 #158 - Why rules must be given as a priori knowledge
Post #136 - How AlphaZero is analogues to evolution
Post #140 - the argument in short form and why the telos answer is irrelevant
Post #153 - my own tentative answer

Hello In Situ.

The idea that there exists a random event, I reject out of hand.

All I have ever observed are events triggering other events and so on.

Computers can never perform a random event.

People never make random choices.

 

[In situ]

The bulldozer is basically a manpower-multiplier. More of the same. But a machine for intellectual feats could do things we cant even imagine... like the way AZ's style of play has taken chess experts totally by surprise.

You are saying bulldozers extends our bodies capacity while machine, like AphaZero, extetends our minds capacity.