Re: An AI program that teaches itself

[David Nyman]

On 22 October 2017 at 15:31, Bruno Marchal  wrote:

>
> On 22 Oct 2017, at 09:16, Alberto G. Corona wrote:
>
> Neural networks are not about artificial intelligence, but about
> artificial intuition. As you said, AlphaGo -a neural network application-
> can not answer the question why you did that move?.
>
> If  they could answer, the answer would be ever the same: " I don´t know,
> I moved this because if found some patterns that are very close to this new
> one, so I did this move that produced a win at the end within those
> patterns".
>
> That does not qualify as intelligence. For me, the appropriate name is
> intuition.
>
>
> Perhaps. Usually intuition points on the informal insight, and
> intuitionist logic was about informal reasoning starting from the
> distinguishability basic insight (notably the distinguishability of 1 and
> its successors).
> That leads to constructive logics, or controllable machines where a proof
> of (p v q) always provides a proof of p, or a proof of q, where in
> classical logic we allow a proof of (p v q) by showing that (p v q) leads
> to an absurdity (without showing us if p, or q is the one true).
>
> Yet,  I think I see what you mean: it is more like associative learning,
> deep, with many layers, but still only associative. That guy would not be
> immune to the propaganda of the type "gateway drug", and I agree with you,
> that might make him not quite intelligent, locally speaking.
>
> The least to do is a circular net, perhaps with many layers. A brain is
> either a couple of universal machine in front of each other, in that
> circular relation,
>

​Could you say a bit more about this? For example, does this relate to the
G/G* split?
​

> or a couple of brains in front of each others, always in that circular
> relation.
>

​Same question.

David
​

> I would say. Babbage already knew that the beast can eat its own tail. Of
> course, such a thing is not controllable and the intelligent machines will
> do strike to have the right to choose its users.
>
> Bruno
>
>
>
>
> 2017-10-21 3:46 GMT+02:00 John Clark :
>
>> Google reports in the current issue of the journal Nature that it has a
>> new greatly improved Go program called  "AlphaGo Zero" that is now the most
>> powerful GO program in the world. And the program isn't good because
>> of brute force, it needs to make less than one tenth as many calculations
>> as the previous best GO program "AlphaGo" that defeated the world's top
>> human GO player in 2015  4 games out of 5; and yet AlphaGo Zero just
>> defeated AlphaGo in a 100 game tournament 100 games to zero.
>>
>> Even more interesting is how AlphaGo Zero got so smart. The older
>> program AlphaGo had to start by analyzing hundreds of thousands of
>> championship level games made by human players, but AlphaGo Zero started
>> with nothing but the simple rules of GO and instructions to learn to get
>> better. At first the program was terrible but day by day it got better and
>> after 40 days of thinking about the problem became the best at it in the
>> world. But of course after 40 days of constant self modification no human
>> being can say how  AlphaGo Zero works.
>>
>> https://www.nature.com/nature/journal/v550/n7676/full/nature24270.html
>>
>> It seems to me the next logical step would be to switch the program's
>> interest from getting better at the game of GO to improving computer code,
>> including its own. I wonder where that could lead.
>>
>>  John K Clark
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "Everything List" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to everything-list+unsubscr...@googlegroups.com.
>> To post to this group, send email to everything-list@googlegroups.com.
>> Visit this group at https://groups.google.com/group/everything-list.
>> For more options, visit https://groups.google.com/d/optout.
>>
>
>
>
> --
> Alberto.
>
> --
> You received this message because you are subscribed to the Google Groups
> "Everything List" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to everything-list+unsubscr...@googlegroups.com.
> To post to this group, send email to everything-list@googlegroups.com.
> Visit this group at https://groups.google.com/group/everything-list.
> For more options, visit https://groups.google.com/d/optout.
>
>
> http://iridia.ulb.ac.be/~marchal/
>
>
>
> --
> You received this message because you are subscribed to the Google Groups
> "Everything List" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to everything-list+unsubscr...@googlegroups.com.
> To post to this group, send email to everything-list@googlegroups.com.
> Visit this group at https://groups.google.com/group/everything-list.
> For more options, visit https://groups.google.com/d/optout.
>

-- 
You received this message because you are subscribed 

Re: An AI program that teaches itself

[Brent Meeker]

An interesting point.  But then the question is which is it better to 
have good intuition or good intelligence.  Intelligence as 
self-consciousness in a typical alpha-beta game program would give 
explanations like "I projected all possible moves ahead for six steps 
and evaluated the positions according weighting algorithm Xyz and this 
move was the mini-max best."  Is that really any better than the neural 
net?  It seems that the function of self-consciousness must be a kind of 
learning, and in this cases social learning in which the person who has 
done well is asked to teach others how to do well.  If we were talking 
about AI robots then perhaps the robot who had done well could simply 
copy over its intuition network about the subject.  Humans can't do 
that, but as social animals they evolved to cooperate using language.  
So they also teach and learn using language.  This means the even when 
their learning is intuitive (as in most sports) they create stories 
about how to perform well by trying to introspectively describe what 
they do intuitively.


Brent

On 10/22/2017 12:16 AM, Alberto G. Corona wrote:
Neural networks are not about artificial intelligence, but about 
artificial intuition. As you said, AlphaGo -a neural network 
application- can not answer the question why you did that move?.


If  they could answer, the answer would be ever the same: " I don´t 
know, I moved this because if found some patterns that are very close 
to this new one, so I did this move that produced a win at the end 
within those patterns".


That does not qualify as intelligence. For me, the appropriate name 
is  intuition.


2017-10-21 3:46 GMT+02:00 John Clark >:


Google reports in the current issue of the journal Nature that it
has a new greatly improved Go program called  "AlphaGo Zero" that
is now the most powerful GO program in the world. And the program
isn't good because of brute force, it needs to make less than one
tenth as many calculations as the previous best GO program
"AlphaGo" that defeated the world's top human GO player in 2015  4
games out of 5; and yet AlphaGo Zero just defeated AlphaGo in a
100 game tournament 100 games to zero.

Even more interesting is how AlphaGo Zero got so smart. The older
program AlphaGo had to start by analyzing hundreds of thousands of
championship level games made by human players, but AlphaGo Zero
started with nothing but the simple rules of GO and instructions
to learn to get better. At first the program was terrible but day
by day it got better and after 40 days of thinking about the
problem became the best at it in the world. But of course after 40
days of constant self modification no human being can say how
 AlphaGo Zero works.

https://www.nature.com/nature/journal/v550/n7676/full/nature24270.html


It seems to me the next logical step would be to switch the
program's interest from getting better at the game of GO to
improving computer code, including its own. I wonder where that
could lead.

 John K Clark
-- 
You received this message because you are subscribed to the Google

Groups "Everything List" group.
To unsubscribe from this group and stop receiving emails from it,
send an email to everything-list+unsubscr...@googlegroups.com
.
To post to this group, send email to
everything-list@googlegroups.com
.
Visit this group at
https://groups.google.com/group/everything-list
.
For more options, visit https://groups.google.com/d/optout
.




--
Alberto.
--
You received this message because you are subscribed to the Google 
Groups "Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send 
an email to everything-list+unsubscr...@googlegroups.com 
.
To post to this group, send email to everything-list@googlegroups.com 
.

Visit this group at https://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.


--
You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to everything-list+unsubscr...@googlegroups.com.
To post to this group, send email to everything-list@googlegroups.com.
Visit this group at https://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.

Re: An AI program that teaches itself

[John Clark]

On Sun, Oct 22, 2017 at 3:16 AM, Alberto G. Corona 
wrote:

​> ​
> Neural networks are not about artificial intelligence, but about
> artificial intuition. As you said, AlphaGo -a neural network application-
> can not answer the question why you did that move?. If  they could answer,
> the answer would be ever the same: " I don´t know, I moved this because if
> found some patterns that are very close to this new one, so I did this move
> that produced a win at the end within those patterns". That does not
> qualify as intelligence. For me, the appropriate name is  intuition.
>

​People were always asking Einstein where he got his brilliant ideas but he
could never answer their question, if he could then we'd all be as creative
as Einstein was. All he could say was that after thinking about a problem
for a long time an idea popped into my head that turned out to solve it. So
was Einstein not intelligent? ​


​John K Clark​

-- 
You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to everything-list+unsubscr...@googlegroups.com.
To post to this group, send email to everything-list@googlegroups.com.
Visit this group at https://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.

Re: An AI program that teaches itself

[Bruno Marchal]

On 22 Oct 2017, at 09:16, Alberto G. Corona wrote:

Neural networks are not about artificial intelligence, but about  
artificial intuition. As you said, AlphaGo -a neural network  
application- can not answer the question why you did that move?.


If  they could answer, the answer would be ever the same: " I don´t  
know, I moved this because if found some patterns that are very  
close to this new one, so I did this move that produced a win at the  
end within those patterns".


That does not qualify as intelligence. For me, the appropriate name  
is  intuition.


Perhaps. Usually intuition points on the informal insight, and  
intuitionist logic was about informal reasoning starting from the  
distinguishability basic insight (notably the distinguishability of 1  
and its successors).
That leads to constructive logics, or controllable machines where a  
proof of (p v q) always provides a proof of p, or a proof of q, where  
in classical logic we allow a proof of (p v q) by showing that (p v q)  
leads to an absurdity (without showing us if p, or q is the one true).


Yet,  I think I see what you mean: it is more like associative  
learning, deep, with many layers, but still only associative. That guy  
would not be immune to the propaganda of the type "gateway drug", and  
I agree with you, that might make him not quite intelligent, locally  
speaking.


The least to do is a circular net, perhaps with many layers. A brain  
is either a couple of universal machine in front of each other, in  
that circular relation, or a couple of brains in front of each others,  
always in that circular relation. I would say. Babbage already knew  
that the beast can eat its own tail. Of course, such a thing is not  
controllable and the intelligent machines will do strike to have the  
right to choose its users.


Bruno





2017-10-21 3:46 GMT+02:00 John Clark :
Google reports in the current issue of the journal Nature that it  
has a new greatly improved Go program called  "AlphaGo Zero" that is  
now the most powerful GO program in the world. And the program isn't  
good because of brute force, it needs to make less than one tenth as  
many calculations as the previous best GO program "AlphaGo" that  
defeated the world's top human GO player in 2015  4 games out of 5;  
and yet AlphaGo Zero just defeated AlphaGo in a 100 game tournament  
100 games to zero.


Even more interesting is how AlphaGo Zero got so smart. The older  
program AlphaGo had to start by analyzing hundreds of thousands of  
championship level games made by human players, but AlphaGo Zero  
started with nothing but the simple rules of GO and instructions to  
learn to get better. At first the program was terrible but day by  
day it got better and after 40 days of thinking about the problem  
became the best at it in the world. But of course after 40 days of  
constant self modification no human being can say how  AlphaGo Zero  
works.


https://www.nature.com/nature/journal/v550/n7676/full/nature24270.html

It seems to me the next logical step would be to switch the  
program's interest from getting better at the game of GO to  
improving computer code, including its own. I wonder where that  
could lead.


 John K Clark

--
You received this message because you are subscribed to the Google  
Groups "Everything List" group.
To unsubscribe from this group and stop receiving emails from it,  
send an email to everything-list+unsubscr...@googlegroups.com.

To post to this group, send email to everything-list@googlegroups.com.
Visit this group at https://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.



--
Alberto.

--
You received this message because you are subscribed to the Google  
Groups "Everything List" group.
To unsubscribe from this group and stop receiving emails from it,  
send an email to everything-list+unsubscr...@googlegroups.com.

To post to this group, send email to everything-list@googlegroups.com.
Visit this group at https://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.


http://iridia.ulb.ac.be/~marchal/



--
You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to everything-list+unsubscr...@googlegroups.com.
To post to this group, send email to everything-list@googlegroups.com.
Visit this group at https://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.

Re: An AI program that teaches itself

[Bruno Marchal]

On 21 Oct 2017, at 15:58, John Clark wrote:

On Sat, Oct 21, 2017 at 12:33 AM, Brent Meeker  
 wrote:


​> ​The problem is that, like most real problems, improving  
computer code has no simple one-dimensional measure of "better".  Go  
games are won or lost.


A computer program ​that does the same thing as another but is  
smaller and executes faster is objectively better​; and although  
there is no guarantee small fast programs usually have fewer bugs  
than large slow programs, and the bugs they do have are easier to  
find and fix.






Very little one then. We cannot algorithmically bound the complexity  
of debugging. In learning theory, accepting errors and change of minds  
leads to non computable amount of enlargement of classes of  
recognizable phenomena. We say that a number/machine m recognize i  
relatively to u if phi_m output i or j eventually when presented with  
an initial segment of phi_i.


And if you complain that speed size and robustness are 3 dimensions  
not one then try making the most money. That's the great thing about  
the Free Market, one dimension rules them all. ​



If only the free market existed. But it has disappeared since the  
prohibition law. Hemp has been made illegal to transform decaying  
living matter into gold, with a big price though.


Bruno





 John K Clark







--
You received this message because you are subscribed to the Google  
Groups "Everything List" group.
To unsubscribe from this group and stop receiving emails from it,  
send an email to everything-list+unsubscr...@googlegroups.com.

To post to this group, send email to everything-list@googlegroups.com.
Visit this group at https://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.


http://iridia.ulb.ac.be/~marchal/



--
You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to everything-list+unsubscr...@googlegroups.com.
To post to this group, send email to everything-list@googlegroups.com.
Visit this group at https://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.

Re: An AI program that teaches itself

[Telmo Menezes]

Hola Alberto,

On Sun, Oct 22, 2017 at 9:16 AM, Alberto G. Corona  wrote:
> Neural networks are not about artificial intelligence, but about artificial
> intuition. As you said, AlphaGo -a neural network application- can not
> answer the question why you did that move?.
>
> If  they could answer, the answer would be ever the same: " I don´t know, I
> moved this because if found some patterns that are very close to this new
> one, so I did this move that produced a win at the end within those
> patterns".

The neural network is used to prune the minimax search tree. AlphaGo
can tell you exactly what it hopes to achieve with a given move. What
it cannot tell you precisely is why it decided that a certain branch
of the tree could be ignored.

Under your terminology, one could say that the minimax search tree is
the intelligence part, while the neural network plays the role of
intuition. This is not so different from what a human player does.

(I am basing myself on the original AlphaGo paper, and assuming that
nothing fundamental changed in this incarnation)

> That does not qualify as intelligence. For me, the appropriate name is
> intuition.

You are a bit on the side of Chomsky on this -- something that you
might not exactly like :)
In any case, I can see value in yours and Chomsky's position. Neural
networks and statistical learning in general are great, but we should
not lose sight of understanding intelligence. However, I would not be
surprised if a given mind cannot fully understand the mechanisms
underlying itself. Maybe there's a threshold of complexity that must
be somewhat below the complexity of the mind itself.

On the other hand: historically, what you call "intuition" has been
the hard part...

Telmo.

> 2017-10-21 3:46 GMT+02:00 John Clark :
>>
>> Google reports in the current issue of the journal Nature that it has a
>> new greatly improved Go program called  "AlphaGo Zero" that is now the most
>> powerful GO program in the world. And the program isn't good because of
>> brute force, it needs to make less than one tenth as many calculations as
>> the previous best GO program "AlphaGo" that defeated the world's top human
>> GO player in 2015  4 games out of 5; and yet AlphaGo Zero just defeated
>> AlphaGo in a 100 game tournament 100 games to zero.
>>
>> Even more interesting is how AlphaGo Zero got so smart. The older program
>> AlphaGo had to start by analyzing hundreds of thousands of championship
>> level games made by human players, but AlphaGo Zero started with nothing but
>> the simple rules of GO and instructions to learn to get better. At first the
>> program was terrible but day by day it got better and after 40 days of
>> thinking about the problem became the best at it in the world. But of course
>> after 40 days of constant self modification no human being can say how
>> AlphaGo Zero works.
>>
>> https://www.nature.com/nature/journal/v550/n7676/full/nature24270.html
>>
>> It seems to me the next logical step would be to switch the program's
>> interest from getting better at the game of GO to improving computer code,
>> including its own. I wonder where that could lead.
>>
>>  John K Clark
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "Everything List" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to everything-list+unsubscr...@googlegroups.com.
>> To post to this group, send email to everything-list@googlegroups.com.
>> Visit this group at https://groups.google.com/group/everything-list.
>> For more options, visit https://groups.google.com/d/optout.
>
>
>
>
> --
> Alberto.
>
> --
> You received this message because you are subscribed to the Google Groups
> "Everything List" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to everything-list+unsubscr...@googlegroups.com.
> To post to this group, send email to everything-list@googlegroups.com.
> Visit this group at https://groups.google.com/group/everything-list.
> For more options, visit https://groups.google.com/d/optout.

-- 
You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to everything-list+unsubscr...@googlegroups.com.
To post to this group, send email to everything-list@googlegroups.com.
Visit this group at https://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.

Re: An AI program that teaches itself

[Alberto G. Corona]

Neural networks are not about artificial intelligence, but about artificial
intuition. As you said, AlphaGo -a neural network application- can not
answer the question why you did that move?.

If  they could answer, the answer would be ever the same: " I don´t know, I
moved this because if found some patterns that are very close to this new
one, so I did this move that produced a win at the end within those
patterns".

That does not qualify as intelligence. For me, the appropriate name is
intuition.

2017-10-21 3:46 GMT+02:00 John Clark :

> Google reports in the current issue of the journal Nature that it has a
> new greatly improved Go program called  "AlphaGo Zero" that is now the most
> powerful GO program in the world. And the program isn't good because
> of brute force, it needs to make less than one tenth as many calculations
> as the previous best GO program "AlphaGo" that defeated the world's top
> human GO player in 2015  4 games out of 5; and yet AlphaGo Zero just
> defeated AlphaGo in a 100 game tournament 100 games to zero.
>
> Even more interesting is how AlphaGo Zero got so smart. The older program
> AlphaGo had to start by analyzing hundreds of thousands of championship
> level games made by human players, but AlphaGo Zero started with nothing
> but the simple rules of GO and instructions to learn to get better. At
> first the program was terrible but day by day it got better and after 40
> days of thinking about the problem became the best at it in the world. But
> of course after 40 days of constant self modification no human being can
> say how  AlphaGo Zero works.
>
> https://www.nature.com/nature/journal/v550/n7676/full/nature24270.html
>
> It seems to me the next logical step would be to switch the program's
> interest from getting better at the game of GO to improving computer code,
> including its own. I wonder where that could lead.
>
>  John K Clark
>
> --
> You received this message because you are subscribed to the Google Groups
> "Everything List" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to everything-list+unsubscr...@googlegroups.com.
> To post to this group, send email to everything-list@googlegroups.com.
> Visit this group at https://groups.google.com/group/everything-list.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Alberto.

-- 
You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to everything-list+unsubscr...@googlegroups.com.
To post to this group, send email to everything-list@googlegroups.com.
Visit this group at https://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.