Re: [Vo]:Article about Artificial Intelligence in NYT

[Jed Rothwell]

NY Times on automation again today:

http://www.nytimes.com/2016/12/21/upshot/the-long-term-jobs-killer-is-not-china-its-automation.html

It's a hot topic.

- Jed

Re: [Vo]:Article about Artificial Intelligence in NYT

[Jed Rothwell]

 wrote:


> I think it's the fact that they have never seen themselves, hence don't
> recognize the image in the mirror.


Which brings to mind the mirror scene in "Duck Soup" -- the funniest
sequence ever filmed, in my opinion.

https://www.youtube.com/watch?v=VKTT-sy0aLg

Re: [Vo]:Article about Artificial Intelligence in NYT

[mixent]

In reply to  Jed Rothwell's message of Sat, 17 Dec 2016 16:22:10 -0500:
Hi,
[snip]
>Perhaps there is something about a reflection that throws off their
>perception. Even cats attack themselves in the mirror, and they are a lot
>smarter than birds.

I think it's the fact that they have never seen themselves, hence don't
recognize the image in the mirror. OTOH they have grown up with their family,
and can easily distinguish them from outsiders. Even young humans only catch on
that they are looking at a reflection when they notice that it mimics all their
actions. BTW this is a reasonable Turing test for AI's.
Regards,

Robin van Spaandonk

http://rvanspaa.freehostia.com/project.html

Re: [Vo]:Article about Artificial Intelligence in NYT

[Che]

On Sat, Dec 17, 2016 at 7:12 PM, Ron Wormus  wrote:

> Robins do this also. Nothing I have found can dissuade them from crashing
> their reflection. I had one persist for over two weeks.



While robins do cooperate while food-gathering, they and bluejays and the
like do not seem to possess what e.g. all the social songbirds seem to
share with Humans: a certain recognizable 'gregariousness'. We Humans
appear to respond strongly to this capability.

Crows/ravens appear to be in a whole other class of intelligence.

Re: [Vo]:Article about Artificial Intelligence in NYT

[Ron Wormus]

Robins do this also. Nothing I have found can dissuade them from crashing 
their reflection. I had one persist for over two weeks.


--On Sunday, December 18, 2016 7:56 AM +1100 mix...@bigpond.com wrote:


In reply to  Jed Rothwell's message of Fri, 16 Dec 2016 22:11:49 -0500:
Hi,
[snip]

My point being, that barely qualifies as conscious. Not as I defined
it: "Awareness of surroundings. Some ability to make choices . . ." It
is more like a set of complicated hard-wired reactions. The cricket
mistakes a plastic object for another cricket. Its perceptions and
awareness of the surroundings are very crude, compared to a bird or
mouse.


Some birds are not much better. I had a little blue jay in mortal combat
with his own reflection in the window during mating season. On the other
side of glass I could sit with my face not six inches away from him and
he still persisted.

Regards,

Robin van Spaandonk

http://rvanspaa.freehostia.com/project.html

Re: [Vo]:Article about Artificial Intelligence in NYT

[Eric Walker]

I wasn't aware of any definitions of consciousness that rule out animals.
But watching enough animal videos on Reddit is sufficient to make one
contemplate vegetarianism. Even less intelligent animals often seem playful
or excited in ways very similar to humans:

https://www.youtube.com/watch?v=cILZ_cB3_so
https://www.youtube.com/watch?v=5f_CxV4eIrU

Eric

Re: [Vo]:Article about Artificial Intelligence in NYT

[a.ashfield]

Jed,
I basically agree.  I think many animals including most mammals and 
birds are conscious,  but there is a wide range of intelligence.
I am baffled by the mysterious, esoteric properties of consciousness 
that academics often apply to it, ruling out all animals.

AA


On 12/16/2016 5:32 PM, Jed Rothwell wrote:

a.ashfield > wrote:

I think it follows we will have a "conscious" computer even
earlier than Kurzweil forecasts.


I think there are various levels of intelligence. Roughly speaking 
here are three points on the spectrum:


1. Thinking. You can make a case that even guppies and earthworms do this.

2. Conscious. Awareness of surroundings. Some ability to make choices, 
rather than purely instinct driving, hard-wired brain functions. I 
expect that mice are conscious. In this article it was estimated that 
present-day artificial intelligence computers have roughly as many 
virtual synapses as a mouse brain has.


3. Sentient, or self-aware. At the lowest level, this means knowing 
that you are an animal and an object in the real world. There is no 
doubt that apes and other intelligent creatures have this. At the zoo 
in Boston, when you take a picture of a chimpanzee with a digital 
camera, it will pose and then demand to see the back of your camera. 
Especially males do this, according to my daughter, who is studying 
biology.


My guess is that computers are somewhere between 1 and 2. They have 
probably not achieved 2 because people who design computers are not 
trying to achieve this at present. Perhaps consciousness will emerge 
on its own as a meta-phenomenon.


There are an infinite number of steps between each level. There are 
various mental achievements. For example, male crickets are capable of 
fighting for domination, which is sophisticated behavior. It is 
impressive for such a small brain. Surely, this is a form of thinking, 
even if it is mainly instinct driven. Unfortunately for the crickets, 
they cannot tell one another apart, and they cannot tell the 
difference between a cricket and a plastic model of one. So, 
naturalists who wanted to give a male cricket an inferiority complex 
engaged in ritual combat with him using a plastic model of a cricket. 
They did this over and over again with the same plastic model. The 
poor guy-cricket did not realize he was fighting the same dummy 
cricket every time. Apparently this sapped his male hormone supply, 
a.k.a. precious bodily fluids.


- Jed

Re: [Vo]:Article about Artificial Intelligence in NYT

[Jed Rothwell]

 wrote:


> Some birds are not much better. I had a little blue jay in mortal combat
> with
> his own reflection in the window during mating season.


A crow used to wake me up in the morning fighting its own reflection. Yet
birds can distinguish other individual birds. It has been shown that
chickens recognize hundreds of other individuals.

Perhaps there is something about a reflection that throws off their
perception. Even cats attack themselves in the mirror, and they are a lot
smarter than birds.

- Jed

Re: [Vo]:Article about Artificial Intelligence in NYT

[mixent]

In reply to  mix...@bigpond.com's message of Sun, 18 Dec 2016 07:56:51 +1100:
Hi,

I think I may have inadvertently given blue jays a bad name. What I saw was
probably a bluebird, not  a blue jay.
Regards,

Robin van Spaandonk

http://rvanspaa.freehostia.com/project.html

Re: [Vo]:Article about Artificial Intelligence in NYT

[mixent]

In reply to  Jed Rothwell's message of Fri, 16 Dec 2016 22:11:49 -0500:
Hi,
[snip]
>My point being, that barely qualifies as conscious. Not as I defined it: 
>"Awareness of surroundings. Some ability to make choices . . ." It is more 
>like a set of complicated hard-wired reactions. The cricket mistakes a plastic 
>object for another cricket. Its perceptions and awareness of the surroundings 
>are very crude, compared to a bird or mouse.

Some birds are not much better. I had a little blue jay in mortal combat with
his own reflection in the window during mating season. On the other side of
glass I could sit with my face not six inches away from him and he still
persisted.

Regards,

Robin van Spaandonk

http://rvanspaa.freehostia.com/project.html

Re: [Vo]:Article about Artificial Intelligence in NYT

[Jed Rothwell]

I wrote:

Unfortunately for the crickets, they cannot tell one another apart, and
> they cannot tell the difference between a cricket and a plastic model of
> one. So, naturalists who wanted to give a male cricket an inferiority
> complex engaged in ritual combat with him using a plastic model of a
> cricket. They did this over and over again with the same plastic model. The
> poor guy-cricket did not realize he was fighting the same dummy cricket
> every time.
>

My point being, that barely qualifies as conscious. Not as I defined it:
"Awareness of surroundings. Some ability to make choices . . ." It is more
like a set of complicated hard-wired reactions. The cricket mistakes a
plastic object for another cricket. Its perceptions and awareness of the
surroundings are very crude, compared to a bird or mouse.

A self-driving car probably has a better "mental model" of the surrounding
environment than the cricket does. Maybe not . . . you can easily fool a
self-driving car with a two-dimensional cutout model of a pedestrian. I saw
a video of that the other day.

The car may not be A.I. based. I wouldn't know.

- Jed

Re: [Vo]:Article about Artificial Intelligence in NYT

[Che]

> There are an infinite number of steps between each level.
>

In the Universe of all material phenomena in general (i.e., so-called
'meta-physics' being pure Idealist wankerism), its development must
necessarily be open-ended and emergent. How could it be any other way.

So there will always be _further_ possible levels beyond any already
achieved. By any entity or collection of entities.

Re: [Vo]:Article about Artificial Intelligence in NYT

[Jed Rothwell]

a.ashfield  wrote:


> I think it follows we will have a "conscious" computer even earlier than
> Kurzweil forecasts.


I think there are various levels of intelligence. Roughly speaking here are
three points on the spectrum:

1. Thinking. You can make a case that even guppies and earthworms do this.

2. Conscious. Awareness of surroundings. Some ability to make choices,
rather than purely instinct driving, hard-wired brain functions. I expect
that mice are conscious. In this article it was estimated that present-day
artificial intelligence computers have roughly as many virtual synapses as
a mouse brain has.

3. Sentient, or self-aware. At the lowest level, this means knowing that
you are an animal and an object in the real world. There is no doubt that
apes and other intelligent creatures have this. At the zoo in Boston, when
you take a picture of a chimpanzee with a digital camera, it will pose and
then demand to see the back of your camera. Especially males do this,
according to my daughter, who is studying biology.

My guess is that computers are somewhere between 1 and 2. They have
probably not achieved 2 because people who design computers are not trying
to achieve this at present. Perhaps consciousness will emerge on its own as
a meta-phenomenon.

There are an infinite number of steps between each level. There are various
mental achievements. For example, male crickets are capable of fighting for
domination, which is sophisticated behavior. It is impressive for such a
small brain. Surely, this is a form of thinking, even if it is mainly
instinct driven. Unfortunately for the crickets, they cannot tell one
another apart, and they cannot tell the difference between a cricket and a
plastic model of one. So, naturalists who wanted to give a male cricket an
inferiority complex engaged in ritual combat with him using a plastic model
of a cricket. They did this over and over again with the same plastic
model. The poor guy-cricket did not realize he was fighting the same dummy
cricket every time. Apparently this sapped his male hormone supply, a.k.a.
precious bodily fluids.

- Jed

Re: [Vo]:Article about Artificial Intelligence in NYT

[a.ashfield]

Jed,
Thank you for posting the link to that most interesting article.
For some years now I had thought that consciousness was just the brain 
making an image of one's surroundings and examining that image.  I see 
now that is unnecessary, that the brain examines the actual image in 
real time and doesn't have to make a new 3D image fifteen times a second 
and look at that.  The basic concept remains though, that consciousness 
is simply the act of examining the image together with inputs from the 
other sensors.
I think it follows we will have a "conscious" computer even earlier than 
Kurzweil forecasts.

AA

On 12/15/2016 5:47 PM, Jed Rothwell wrote:
There was a long article about artificial intelligence (AI) in the New 
York Times:


http://www.nytimes.com/2016/12/14/magazine/the-great-ai-awakening.html

It was pretty good, with some technical detail. You will find more 
detail in recent Sci. Am. article by two of the leading people in the 
field. I can't find it on line . . .


There has been an important breakthrough with neural networks. They 
have been around for decades, going back to the 1950s I think. The 
difference is they are now many orders of magnitude larger, and they 
are multi-level, with the output of one network connected to the input 
of another.


This was the technique that led to a computer beating the world 
champion in go. The NYT reports that Google has applied this to their 
translation software, resulting in dramatic improvements. In a few 
months, the quality of the translations improved as much as it did in 
10 years with the older technology. The article quotes an example. 
This sentence in Spanish by Borges:


Uno no es lo que es por lo que escribe, sino por lo que ha leído.

The old Google translate system rendered this:

One is not what is for what he writes, but for what he has read.

The new one:

You are not what you write, but what you have read.


I ran some Japanese and some English text through the latest Google 
translate. This is mainly text that I translated myself. The new 
Google translate is remarkable. A little unnerving. Because, you see, 
if you run some of my translated essays you will see that I took 
liberties, adding information I thought would help a native speaker 
understand. These are not literal or exact translations. Since I wrote 
the original text myself, I am allowed to to that. But, I also did it 
with Mike McKubre's paper. I am a little worried that someone may call 
me out on it! Sooner or later, Google's computers will be getting in 
touch with me, calling me out. Google sells a gadget that sits in the 
room listening to your conversations, awaiting your commands, the 
Google Home:


https://madeby.google.com/home/

I can see the day coming when the Google Home speaker will blare out: 
"ROTHWELL! Get over here. What is the meaning of this?!? McKubre wrote 
'I was tasked' and you have it: 'the conference organizers asked me to 
. . .' We are now in the process of reviewing all of your work going 
back to 1998, which will henceforth be called Calendar Year 1 of Our 
Lord Google."


http://lenr-canr.org/acrobat/McKubreMCHcoldfusionb.pdf

http://lenr-canr.org/acrobat/McKubreMCHcoldfusionc.pdf

- Jed

Re: [Vo]:Article about Artificial Intelligence in NYT

[Frank Znidarsic]

-Original Message-
From: Jack Cole <jcol...@gmail.com>
To: vortex-l <vortex-l@eskimo.com>
Sent: Fri, Dec 16, 2016 7:10 am
Subject: Re: [Vo]:Article about Artificial Intelligence in NYT



It is rather remarkable.  A recent story along these lines is that it can even 
translate between two untrained languages.


For example, if the system has been trained to translate between English and 
Spanish and English and Portuguese, then it can reasonably translate between 
Spanish and Portuguese even though it has not been trained.


https://techcrunch.com/2016/11/22/googles-ai-translation-tool-seems-to-have-invented-its-own-secret-internal-language/



On Thu, Dec 15, 2016 at 4:48 PM Jed Rothwell <jedrothw...@gmail.com> wrote:


There was a long article about artificial intelligence (AI) in the New York 
Times:


http://www.nytimes.com/2016/12/14/magazine/the-great-ai-awakening.html



It was pretty good, with some technical detail. You will find more detail in 
recent Sci. Am. article by two of the leading people in the field. I can't find 
it on line . . .


There has been an important breakthrough with neural networks. They have been 
around for decades, going back to the 1950s I think. The difference is they are 
now many orders of magnitude larger, and they are multi-level, with the output 
of one network connected to the input of another.


This was the technique that led to a computer beating the world champion in go. 
The NYT reports that Google has applied this to their translation software, 
resulting in dramatic improvements. In a few months, the quality of the 
translations improved as much as it did in 10 years with the older technology. 
The article quotes an example. This sentence in Spanish by Borges:


Uno no es lo que es por lo que escribe, sino por lo que ha leído.


The old Google translate system rendered this:


One is not what is for what he writes, but for what he has read.


The new one:


You are not what you write, but what you have read.





I ran some Japanese and some English text through the latest Google translate. 
This is mainly text that I translated myself. The new Google translate is 
remarkable. A little unnerving. Because, you see, if you run some of my 
translated essays you will see that I took liberties, adding information I 
thought would help a native speaker understand. These are not literal or exact 
translations. Since I wrote the original text myself, I am allowed to to that. 
But, I also did it with Mike McKubre's paper. I am a little worried that 
someone may call me out on it! Sooner or later, Google's computers will be 
getting in touch with me, calling me out. Google sells a gadget that sits in 
the room listening to your conversations, awaiting your commands, the Google 
Home:


https://madeby.google.com/home/



I can see the day coming when the Google Home speaker will blare out: 
"ROTHWELL! Get over here. What is the meaning of this?!? McKubre wrote 'I was 
tasked' and you have it: 'the conference organizers asked me to . . .' We are 
now in the process of reviewing all of your work going back to 1998, which will 
henceforth be called Calendar Year 1 of Our Lord Google."


http://lenr-canr.org/acrobat/McKubreMCHcoldfusionb.pdf



http://lenr-canr.org/acrobat/McKubreMCHcoldfusionc.pdf




- Jed

Re: [Vo]:Article about Artificial Intelligence in NYT

[Jed Rothwell]

Jack Cole  wrote:


> . . . A recent story along these lines is that it can even translate
> between two untrained languages.
>
> For example, if the system has been trained to translate between English
> and Spanish and English and Portuguese, then it can reasonably translate
> between Spanish and Portuguese even though it has not been trained.
>
> https://techcrunch.com/2016/11/22/googles-ai-translation-
> tool-seems-to-have-invented-its-own-secret-internal-language/
>

That's amazing. But it is not the best test they could have run. They
translated from Japanese into Korean. Japanese and Korean are similar in
many ways. This is like going from Spanish into Italian. They should try it
from Japanese to Russian, or Japanese to Navajo.

- Jed

Re: [Vo]:Article about Artificial Intelligence in NYT

[Jack Cole]

It is rather remarkable.  A recent story along these lines is that it can
even translate between two untrained languages.

For example, if the system has been trained to translate between English
and Spanish and English and Portuguese, then it can reasonably translate
between Spanish and Portuguese even though it has not been trained.

https://techcrunch.com/2016/11/22/googles-ai-translation-tool-seems-to-have-invented-its-own-secret-internal-language/

On Thu, Dec 15, 2016 at 4:48 PM Jed Rothwell  wrote:

> There was a long article about artificial intelligence (AI) in the New
> York Times:
>
> http://www.nytimes.com/2016/12/14/magazine/the-great-ai-awakening.html
>
> It was pretty good, with some technical detail. You will find more detail
> in recent Sci. Am. article by two of the leading people in the field. I
> can't find it on line . . .
>
> There has been an important breakthrough with neural networks. They have
> been around for decades, going back to the 1950s I think. The difference is
> they are now many orders of magnitude larger, and they are multi-level,
> with the output of one network connected to the input of another.
>
> This was the technique that led to a computer beating the world champion
> in go. The NYT reports that Google has applied this to their translation
> software, resulting in dramatic improvements. In a few months, the quality
> of the translations improved as much as it did in 10 years with the older
> technology. The article quotes an example. This sentence in Spanish by
> Borges:
>
> Uno no es lo que es por lo que escribe, sino por lo que ha leído.
>
> The old Google translate system rendered this:
>
> One is not what is for what he writes, but for what he has read.
>
> The new one:
>
> You are not what you write, but what you have read.
>
>
> I ran some Japanese and some English text through the latest Google
> translate. This is mainly text that I translated myself. The new Google
> translate is remarkable. A little unnerving. Because, you see, if you run
> some of my translated essays you will see that I took liberties, adding
> information I thought would help a native speaker understand. These are not
> literal or exact translations. Since I wrote the original text myself, I am
> allowed to to that. But, I also did it with Mike McKubre's paper. I am a
> little worried that someone may call me out on it! Sooner or later,
> Google's computers will be getting in touch with me, calling me out. Google
> sells a gadget that sits in the room listening to your conversations,
> awaiting your commands, the Google Home:
>
> https://madeby.google.com/home/
>
> I can see the day coming when the Google Home speaker will blare out:
> "ROTHWELL! Get over here. What is the meaning of this?!? McKubre wrote 'I
> was tasked' and you have it: 'the conference organizers asked me to . . .'
> We are now in the process of reviewing all of your work going back to 1998,
> which will henceforth be called Calendar Year 1 of Our Lord Google."
>
> http://lenr-canr.org/acrobat/McKubreMCHcoldfusionb.pdf
>
> http://lenr-canr.org/acrobat/McKubreMCHcoldfusionc.pdf
>
> - Jed
>
>

[Vo]:Article about Artificial Intelligence in NYT

[Jed Rothwell]

There was a long article about artificial intelligence (AI) in the New York
Times:

http://www.nytimes.com/2016/12/14/magazine/the-great-ai-awakening.html

It was pretty good, with some technical detail. You will find more detail
in recent Sci. Am. article by two of the leading people in the field. I
can't find it on line . . .

There has been an important breakthrough with neural networks. They have
been around for decades, going back to the 1950s I think. The difference is
they are now many orders of magnitude larger, and they are multi-level,
with the output of one network connected to the input of another.

This was the technique that led to a computer beating the world champion in
go. The NYT reports that Google has applied this to their translation
software, resulting in dramatic improvements. In a few months, the quality
of the translations improved as much as it did in 10 years with the older
technology. The article quotes an example. This sentence in Spanish by
Borges:

Uno no es lo que es por lo que escribe, sino por lo que ha leído.

The old Google translate system rendered this:

One is not what is for what he writes, but for what he has read.

The new one:

You are not what you write, but what you have read.


I ran some Japanese and some English text through the latest Google
translate. This is mainly text that I translated myself. The new Google
translate is remarkable. A little unnerving. Because, you see, if you run
some of my translated essays you will see that I took liberties, adding
information I thought would help a native speaker understand. These are not
literal or exact translations. Since I wrote the original text myself, I am
allowed to to that. But, I also did it with Mike McKubre's paper. I am a
little worried that someone may call me out on it! Sooner or later,
Google's computers will be getting in touch with me, calling me out. Google
sells a gadget that sits in the room listening to your conversations,
awaiting your commands, the Google Home:

https://madeby.google.com/home/

I can see the day coming when the Google Home speaker will blare out:
"ROTHWELL! Get over here. What is the meaning of this?!? McKubre wrote 'I
was tasked' and you have it: 'the conference organizers asked me to . . .'
We are now in the process of reviewing all of your work going back to 1998,
which will henceforth be called Calendar Year 1 of Our Lord Google."

http://lenr-canr.org/acrobat/McKubreMCHcoldfusionb.pdf

http://lenr-canr.org/acrobat/McKubreMCHcoldfusionc.pdf

- Jed