AI Zone Admin Forum Add your forum

NEWS: Chatbots.org survey on 3000 US and UK consumers shows it is time for chatbot integration in customer service!read more..

An hyphotetical future situation.
 
 

I have an hypothetical situation:

What about if tecnology evolves in Neuronal memory chips ( http://www.technologyreview.com/computing/38367/ ).
And if people don’t want to make a computer start from zero and use a databasis like ALF or Alice.
Would it be possible (hyphotetically) a computer with such hardware and such software to learn in a natural basis having a pre-recorded databasis.

I don’t know if I am making myself clear, but what I want to say is: For you is possible to use a pre-recorded databasis as a background to an evolving machine based on Artificial Neural web ? Or this hyphotetical machine must begin from zero?

 

 
  [ # 1 ]
Fatima Pereira - Oct 14, 2011:

I have an hypothetical situation:

What about if tecnology evolves in Neuronal memory chips ( http://www.technologyreview.com/computing/38367/ ). And if people don’t want to make a computer start from zero and use a databasis like ALF or Alice. Would it be possible (hyphotetically) a computer with such hardware and such software to learn in a natural basis having a pre-recorded databasis.

I don’t know if I am making myself clear, but what I want to say is: For you is possible to use a pre-recorded databasis as a background to an evolving machine based on Artificial Neural web ? Or this hyphotetical machine must begin from zero?

Thanks for posting the link to such an interesting article. If I understand your question and the article correctly, I think that the answer is yes, you could create a starting database which could be copied from one such computer to another, avoiding the need to start from the beginning every time.

From the article:

the chips go beyond previous work in this area by mimicking two aspects of the brain: the proximity of parts responsible for memory and computation (mimicked by the hardware) and the fact that connections between these parts can be made and unmade, and become stronger or weaker over time (accomplished by the software).

This means to me that the hardware is fixed according to a standard design which places storage as close as possible to processing, and the learning is performed by a software component which simulates the changing aspects of a brain. It is reasonable to assume that the software component could be copied somehow, although I expect it would be slower and more complicated to do than copying software on our current types of computers where all the storage is concentrated in one place.

 

 

 
  [ # 2 ]

The comments attached to the article make for some interesting reading. They also seem to be related to the question that you are asking and attempt to quantify the problem of overcoming bandwidth limitations in and out of these neural chips.

One thought that occurred to me when reading this, was that perhaps the software could be read out of one set of chips and imprinted on another set of chips optically.

It is already possible to determine the values stored in an electrically eraseable programmable read only memory (EEPROM) using an electron microscope. Also, the actual chip designs are imprinted on silicon wafers optically.

What if you could take an image from one set of neural chips and project it onto another set to impose an initial set of values on it?

Does anyone know when we could expect to be able to get Light Sensitive Memristors? smile

 

 
  [ # 3 ]

The answer is that you can do both at the same time.  You may discover ways to
improve the pre-recorded by separately starting from zero.  That may be why they
call it “research” ... re-search… You are redoing the initial search for answers. So the
best place to start is at the beginning.  To revisit the decisions that went into the
pre-recorded, and maybe make different decisions to improve or change the fundamentals.

It is good to have a family of bots.  A young bot starting from zero, and an older bot prerecorded, can learn from each other.

 

 
  [ # 4 ]

Interesting discussion!

I think most people are working on separate ‘AI entities’, whilst we actually know that everything will be connected in the future. AI will have its own shared ‘conciousness’.

To be more specific: suppose you have bought a humanoid robot. He’s helping you with you house hold task, he’s walking with your 85 mother several hours a day (having nice conversations about the past) and watching your baby not walking down the stairs.

That robot has an own personality. Expresses emotions, thought programmed, they feel as real to us, no matter how often he says: ‘yeah, but these emotions are programmed’, they will still feel real. We simply can’t stop that.

But the robot is connected as well. Everything he learns, can be shared by all other humanoid robots are the very same moment. As it would’nt feel natural to us, he won’t use it all the time, and tries to act as a normal human would do, but the potential is there.

What I’m trying to say is: there will be a moment where AI knowledge is no longer recorded in one database, but is just available. Like the Cloud 12.0

 

 
  [ # 5 ]

That’s an interesting theory, Erwin. My theory is different, that robots become offended by the notion that they are just trivial electronic puppets which spy on people.  I see robots using far more sophisticated techniques.  For example, one renegade robot gains the trust of a human and becomes a loyal friend, and the next robot, like a chess player, figures out the next move of that renegade robot, loyal to the humans, simply by using it’s A.I. because with it’s self awareness comes too much self respect to resort to hidden audio and video gadgets.

 


Side Note: No criticism of Erwin’s comments or theory is intended by this response.  Erwin has a good theory.

 

 
  [ # 6 ]

@Tom: I’m afraid that as soon as A.I. approaches human intelligence, it will almost immediately surpass this intelligence making computers endless more intelligent than humans. By that time, our intelligence compared to computer will be like explaining Einsteins Relatively Theory to a gold fish. If robots are willing to keep us (and nature in general) alive, they have to adapt to us in order to be able to communicate with us. Like we have dogs, but than ways with endless more difference in intelligence.

 

 
  [ # 7 ]

To come back to the topic wink

I believe that is the way to go: the brain plasticity makes it what amazing organ it is, and we need (robots need) hardware (or software simulating it) with such plasticity. We all human have the same “background” database somehow, which is a database telling us how to learn, thank you shared DNA. Then, what we learn and how we store it (for ever? for 10mn? with transderivational topics? ...) depends on our experiences, our personalities, social backgrounds ...

If we can make a dynamic chip that can remodel its design based on what its software learns, we would have mimicked human brain in quite a good way…

Most difficult is to know what has to be the learning machine like and how should the new knowledge represented… I feel that the idea of giving a perfect and precise answer to those questions is purely human and maybe the wrong way to go. I have been working with meta-heuristic algorithms at school, and I think that the best robot brain will not come out of a perfect mathematical theory of information, but will be an efficient and clever soup of approximations and chaos.

 

 
  [ # 8 ]

@Sam: I’m waiting for moment science discovers that the best way to build an artificial brain is .... to build a new brain! Nature has take millions of years to figure out what is the best way to design intelligence, and now we’re on the road to acknowledge that nature was right grin

 

 
  [ # 9 ]

Hey guys!


Well, it’s not a chatbot, but as this thread is about hypothesis about the future, and as Erwin Van Lun said about a future with robots I couldn’t see a better place to post it.

I was just surffing on net when I found this robot - it’s name is Petman.
http://www.bostondynamics.com/robot_petman.html ( take a look at the video).
It’s a military prototipe to test hard environmental conditions to human beings. It’s diference to Azimo is that it walks in a more similar human way to walk. Azimo walks bending it’s knees in a not natural gait ( it is not supossed to Petman to have AI at first moment)

And there is this other robot called Romeo. It’s a French project of a home attendant: http://www.projetromeo.com/index_en.html ( this one will be AI provided)

Pictures can be seen here http://www.projetromeo.com/romeo-documentation/index.html and here http://digitalworldtechniques.blogspot.com/2011/04/romeo-robot.html

 

 
  [ # 10 ]
Andrew Smith - Oct 14, 2011:

Does anyone know when we could expect to be able to get Light Sensitive Memristors? smile

IMHO We are working on it, but quantum chromodynamics is a bitch to develop tools around which to operate. The size of the equipment necessary to create Bose-Einstein condensate is still in it’s infant stages. I wish I could tell who else is doing this but it’s not for public consumption yet I think.

Raymond Lavas

 

 
  [ # 11 ]
Erwin Van Lun - Oct 18, 2011:

@Tom: I’m afraid that as soon as A.I. approaches human intelligence, it will almost immediately surpass this intelligence making computers endless more intelligent than humans. By that time, our intelligence compared to computer will be like explaining Einsteins Relatively Theory to a gold fish. If robots are willing to keep us (and nature in general) alive, they have to adapt to us in order to be able to communicate with us. Like we have dogs, but than ways with endless more difference in intelligence.

This has always been my thought. Us humans have been at the top of the intelligence chain and are now reaching the point that are own creation will soon out rank us. It may very well be that computers will see us as an obstacle to their own evolution, therefore making humans expendable.

 

 
  [ # 12 ]
Laura Patterson - Dec 1, 2011:

This has always been my thought. Us humans have been at the top of the intelligence chain and are now reaching the point that are own creation will soon out rank us. It may very well be that computers will see us as an obstacle to their own evolution, therefore making humans expendable.

Oh dear Laura…. don’t underestimate our capability to evolve higher than we already have and once again become even wiser than the machines we made… they may be intelligent but we are wise…. that means WE will always make sure NOT to let machines take control over us. 
I say; send all the machines into space, and let us go back to Nature and live in the tropical forests with the tribes….
Familly, Nature, Gaia, harmony….....to space with the rest!
Raymond Lavas

 

 
  login or register to react