World's top Go player loses second match of five against Google AI

[RichardChad]

Kurros,

Please tell me what I'm getting wrong. Help me.
(And don't just say...oh you don't understand....read moar..it's really special...neural networks...fuzzy logic...algorithms...it's all in the jargon.)

To what extent does the program have the 'freedom' to make its own choice of moves entirely free of compulsion?

Basic program - IF this, then do THAT. IF this, then do THAT. IF this, then do THAT....etc.  etc. And do it as fast as the (man made) hardware allows.

The (human) programmer can theoretically write a program that includes every known combination and 'best moves' based on game outcome statistics from past games.

The programmer can also write into the game software a "coin toss" random decision function for situations Wher two competing "best moves" have an equiprobable chance of leading to a win.

So in what sense does the AI game program do something which is NOT expected?

There are no if statements in the model, and the programmer does not give it any information about the game. The program learns to play the same way humans do: simply by playing over and over and seeing what works and what doesn't.

Equivocation, there may be no "IF" statements (depending on the language used), but there are most certainly conditionals.

The program simply assembles more information upon which conditionals work. Perfectly determined, perfectly expected. You wont attempt to argue that point.

This is part of how humans learn, but it doesn't have the ability to freely choose that humans have.

[Trinity]

ADS meeting AlphaGo:

http://webcolleges.uva.nl/Mediasite/Play/d9d5c6e919034305ab8cad4e1d430a0f1d

I haven't watched the video myself, but I assume others here may be interested to see this college presentation.

[aleph naught]

This is part of how humans learn, but it doesn't have the ability to freely choose that humans have.

No one ever said it did. We're talking about artificial intelligence, not free will.

[RichardChad]

This is part of how humans learn, but it doesn't have the ability to freely choose that humans have.

No one ever said it did. We're talking about artificial intelligence, not free will.

I'm the one pointing out computers don't have LFW, you're the one responding to "To what extent does the program have the 'freedom' to make its own choice of moves entirely free of compulsion?"

always the trick with you eh? : - )

[TheBigOhMan]

This is part of how humans learn, but it doesn't have the ability to freely choose that humans have.

No one ever said it did. We're talking about artificial intelligence, not free will.

Ive always suspected that the reason why no machine will ever have free will is because free will itself is impossible.

[Trinity]

Ive always suspected that the reason why no machine will ever have free will is because free will itself is impossible.

That doesn't necessarily follow though.

[Lion IRC]

...There are no if statements in the model, and the programmer does not give it any information about the game. The program learns to play the same way humans do: simply by playing over and over and seeing what works and what doesn't.

All I'm seeing is a program which stores game statistics and plays the odds based on those stats.
It doesn't 'teach itself' to play.

It records past game data in its memory and - surprise, surprise - the more games it plays, the more statistical data it accumulates therefore increasing the odds of a win.

It's a brute force winning machine.

The program played 'itself' and other game software programs over and over and over (at high speed) accumulating statistical correlations between moves and the statistical probability of a win/loss. That's thousands of games!

OF COURSE it's going to beat a single human who can never memorise all those game statistics.

[igr]

Even though this is about free will, what does that really mean in practise?  Maybe it is about what we take into account when making a decision, and how much weight to give to each thing we take into account.  AI could do this part, with a bit of random chucked in.  Maybe that would be indistinguishable from an actual human in making a decision.

[TheBigOhMan]

Ive always suspected that the reason why no machine will ever have free will is because free will itself is impossible.

That doesn't necessarily follow though.

It was just a suggestion.

[Rostos]

This is part of how humans learn, but it doesn't have the ability to freely choose that humans have.

No one ever said it did. We're talking about artificial intelligence, not free will.

Ive always suspected that the reason why no machine will ever have free will is because free will itself is impossible.

Were you free to come to that conclusion?

[TheBigOhMan]

This is part of how humans learn, but it doesn't have the ability to freely choose that humans have.

No one ever said it did. We're talking about artificial intelligence, not free will.

Ive always suspected that the reason why no machine will ever have free will is because free will itself is impossible.

Were you free to come to that conclusion?

Probably not,

[Rostos]

This is part of how humans learn, but it doesn't have the ability to freely choose that humans have.

No one ever said it did. We're talking about artificial intelligence, not free will.

Ive always suspected that the reason why no machine will ever have free will is because free will itself is impossible.

Were you free to come to that conclusion?

Probably not,

Therefore you cannot rationally affirm it

[wonderer]

This is part of how humans learn, but it doesn't have the ability to freely choose that humans have.

No one ever said it did. We're talking about artificial intelligence, not free will.

Ive always suspected that the reason why no machine will ever have free will is because free will itself is impossible.

Were you free to come to that conclusion?

Probably not,

Therefore you cannot rationally affirm it

I'd suggest TBOM can rationally affirm it based on the frequency with which the deterministic system between TBOM ears comes to correct conclusions, as compared to the deterministic systems between the ears of others/most people.

On the other hand, some people have long histories of demonstrating that they are not free to come to a correct conclusion. (At least in some regards.)

[OrdinaryClay]

Games are won by massive search trees. Computers will always be able to do this better. No news. No surprise.

[kurros]

...There are no if statements in the model, and the programmer does not give it any information about the game. The program learns to play the same way humans do: simply by playing over and over and seeing what works and what doesn't.

All I'm seeing is a program which stores game statistics and plays the odds based on those stats.
It doesn't 'teach itself' to play.

It records past game data in its memory and - surprise, surprise - the more games it plays, the more statistical data it accumulates therefore increasing the odds of a win.

It's a brute force winning machine.

The program played 'itself' and other game software programs over and over and over (at high speed) accumulating statistical correlations between moves and the statistical probability of a win/loss. That's thousands of games!

OF COURSE it's going to beat a single human who can never memorise all those game statistics.

It doesn't memorise all those game statistics. As we keep saying, what you are describing is simply not how AlphaGo works. Learning algorithms do not work by just storing all the past game data. Same as the Google image search AI does not store all the pictures of cats it looks at. It uses those pictures to update neural network connections and weights, so that the network becomes better and better at recognising cats. It is really much more like what the human brain does than you seem to realise.