"Her"

[Magical Realist]

I got to see "Her" this weekend, which is the new Spike Jonze movie about a man in the not-too-remote future who falls in love with an artificially intelligent operating system. The film is excellent, insightful, provocative and just plain beautiful to look at in it's ochre filtered tones. At one point the OS, played by the voiceover of Scarlett Johansen, wonders if her feelings are artificial since they are merely part of her programming. But aren't ALL feelings part of the program? Whose to say human feelings aren't real just because they are hardwired into us thru millions of years of evolution? In any case, check this movie out if you get a chance. It'll make you think about the distinction, perhaps itself dubious, between human and machine consciousness.

 

[Buddha12]

I thought AI was very good at looking at that. It showed many aspects of human/ machine interaction and what can happen.

 

[kwhilborn]

Computer programming is extremely simple. All a computer can ever do is compare 2 values at a time and turn on lights and sounds (etc.) based upon those decisions.

It is amazing how such a simple thing as comparing 2 values can be manipulated into fooling users into recognizing what might appear to be intelligence from our computers. It is all illusion.

A person would need to be a non programming moron to ever think a Program will think.

People discuss Bio Brains that learn like we do as a POSSIBLE future. This would not really be a machine though would it.

Also programming is vital for computer use and a bio brain would not be programmable as much as taught so I doubt heavy funding for such a thing would ever prove practical.

AI is just a nice fantasy.

 

[Magical Realist]

It is amazing how such a simple thing as comparing 2 values can be manipulated into fooling users into recognizing what might appear to be intelligence from our computers. It is all illusion.

Whose to say that conscious intelligence isn't an emergent phenomena that rises out of its functioning components? I mean people are still struggling to understand how human consciousness can emerge from a bunch of firing synapses. What if the same could be accomplished on the computer level, particularly with qubits in a quantum computer?

 

[kwhilborn]

@ MR,

If you are suggesting that anything can be intelligent (such as a coffeemaker) then that is a point. A coffeemaker must decide whether to turn on at a given time or if a button is pressed. This comparison is similar to any computer software in existence.

Despite the appearance of intelligence, software can only do comparisons and turn on blinking lights and sound.

Even a quantum computer must work this way or it would be useless in present day applications. When you set your alarm the processor does not DECIDE whether or not to wake you up when you asked. It follows instructions.

A computer program is constantly comparing 2 values. This is in no way the same as thinking.

 

[billvon]

Computer programming is extremely simple. All a computer can ever do is compare 2 values at a time and turn on lights and sounds (etc.) based upon those decisions.

All our brain's neurons can do is fire or not. Those are the only two things a neuron can ever do. Yet by combining a lot of them we get what we call thinking, consciousness, awareness, sentience etc.

 

[lightgigantic]

All our brain's neurons can do is fire or not. Those are the only two things a neuron can ever do. Yet by combining a lot of them we get what we call thinking, consciousness, awareness, sentience etc.

This is simply a particular reductionist theory of thinking, consciousness, etc.

There is no working model of how they "combine" to arrive at such a point.
In fact there are even suggestions that drawing an analogy between human consciousness and a computer simply misses the point of consciousness being the unifying element that exists beyond mere yes/no algorithms.

 

[billvon]

This is simply a particular reductionist theory of thinking, consciousness, etc.

There is no working model of how they "combine" to arrive at such a point.

Agreed. But we do know that elements such as neurons - as simple as the logic blocks of a computer - can combine to produce consciousness. Thus there is nothing preventing a computer doing the same, given the same combinations.

 

[Buddha12]

Douglas Hofstadter actually tries to figure out how to program AI in his writings.

Douglas Richard Hofstadter (born February 15, 1945) is an American professor of cognitive science whose research focuses on the sense of "I"

A few of his books:

Gödel, Escher, Bach: An Eternal Golden Braid

Masters of Deception: Escher, Dalí & the Artists of Optical Illusion

I Am a Strange Loop

http://www.google.com/url?sa=t&rct=..._pNstleMWEZpcSu8TtFSsYA&bvm=bv.59568121,d.b2I

 

[kwhilborn]

@ Billvon,

The Brains Neurons are not just on off switches. If we compare it to a light switch then the wiring within the walls is also changing and sometimes the electricity is just not wired to it, so the switch would be useless.

Perhaps you would need to spend months writing code before you realize computers are just over-sized calculators.

 

[billvon]

The Brains Neurons are not just on off switches.

Didn't say they were. But all they can ever do is fire or not fire - just as all a computer's transistors can ever do is turn on and off. But combine enough of them and you have intelligence.

If we compare it to a light switch then the wiring within the walls is also changing and sometimes the electricity is just not wired to it, so the switch would be useless.

And if you compare it to a computer, the instructions for driving that switch are also changing constantly.

[quote[Perhaps you would need to spend months writing code before you realize computers are just over-sized calculators.[/QUOTE]

I've spent decades doing just that. There is nothing that stops a computer from doing exactly what a brain does; it's just a matter of programming.

(Or to put it another way - your brain is nothing but an oversized calculator either, it's just running better software.)

 

[kwhilborn]

@ Billvon,

But combine enough of them and you have intelligence.

We have some pretty big Supercomputers in the world. Are you suggesting they are intelligent?

A computer can only make one decision at a time, while Neurons run in a magnitude of directions simultaneously. Like a combination of programs operating in unison.

However. I can see where we are getting into opinion now and you do not seem rational about this.

I thought you were more Skeptic Oriented, but you speak about AI as if it is possible.

Since you claim knowledge of programming (seems unlikely given your stance), tell me how a computer could ever be programmed to imagine or invent a solution not in its programming.

Intelligence should be able to do these things.

There are only 2 types of computer coding. Comparing 2 things, and directing an instruction. Brains are not Binary. We can alter our thoughts midstream with a distraction. How do you distract a program?

Anyways. If it helps you sleep at night believe what you like. It would be good sci fi.

 

[wynn]

But we do know that elements such as neurons - as simple as the logic blocks of a computer - can combine to produce consciousness.

No, we don't know that. Some people just define it as such.

 

[wynn]

I thought you were more Skeptic Oriented, but you speak about AI as if it is possible.

Oh, dumbing things down so that they seem manageable can make anything seem possible, and acceptable.

 

[wynn]

I thought AI was very good at looking at that. It showed many aspects of human/ machine interaction and what can happen.

And yet he repeated for thousands of years -
Please, Blue Fairy, make me a real live boy!

 

[Kittamaru]

@ Billvon,



We have some pretty big Supercomputers in the world. Are you suggesting they are intelligent?

A computer can only make one decision at a time, while Neurons run in a magnitude of directions simultaneously. Like a combination of programs operating in unison.

You are comparing calculating capacity to intelligence, which are two very different things. Intelligence/Sentience suggests an awareness of self and self-identity, something computers lack... for now.

As far as a computer only making one decision at a time, that is not really true. They can process many, many simultaneous calculations at once, thanks to multi-threading, multi-core processing, multi-core rendering, and multiple discrete processing cores (CPU, GPU, Physics Processor, etc). It isn't so much about the power of a single core (much like it isn't about the power of a single neural synapse, thank goodness)... it's about the ability of the machine as a whole to co-process. The brain is, essentially, a giant cell-processing matrix... which is good because the processing power of a single neuron is incredibly low. What makes it so capable, though, is that a neuron can help process a colour identity one moment, then help process a muscle motor control command another, and then go on to help process and store a memory of a smell another time. They are incredibly adaptable and that's the part that is so incredibly difficult to program into a chip; as it is, computers can only do what we tell them because trying to tell them to "improve themselves" tends to lead to bad consequences.

 

[Magical Realist]

Deep Learning: A revolution in AI

http://www.nytimes.com/2012/11/24/s...-a-part-of-artificial-intelligence.html?_r=1&

 

[Janus58]

There will always be those that, no matter how sophisticated computers become, will claim that they are not truly intelligent, and that they can only mimic intelligence. But I think that this can be a little like saying that because planes don't travel through the air in the same way as birds do, what they do isn't really "flight".

But there is a reason why we don't make planes that fly like birds do. It's because that, while the way they fly works well for birds, it doesn't work for what we want to use flight for, which is moving much heavier items. Planes don't fly "like the birds", but they do fly in a way that is much more useful for our purposes.

AI, may just turn out the same way. It may turn out to be possible to make a computer that thinks like a human, but then such a computer could be then hampered by the same foibles that plague human thinking. So maybe, for practical purposes, a computer that "thinks like a human" might not be that useful. Then you really just might end up with something that "makes mistakes 1000 times faster than we do".

So it maybe that AI will ultimately be something different. It can do many of the things we can, but without "thinking" like we do. But then it becomes a little harder to decide if they are really "thinking". With planes, it is pretty obvious that while they don't fly like birds, they do fly, but with thought, that might not be so easy to see. Would we be able to recognize a type of intelligence not like our own as intelligence?

And even if the consensus ends up accepting this new type of intelligence as being true intelligence, there will be those who will resist the notion.

And then we get into the whole moral/ethical issues. At what point along the spectrum between what we have today and what develops in the future could we develop something that could be considered a being in its own right? Will we see the formation of a "AI rights" movement? With one side arguing for them and the other arguing that they are just machines and can't really think.

Anyway, just some musings of mine on the subject.

 

[billvon]

We have some pretty big Supercomputers in the world. Are you suggesting they are intelligent?

Nope. They don't have the right software.

A computer can only make one decision at a time, while Neurons run in a magnitude of directions simultaneously.

A neuron can only make one decision - fire or not. A lot of them together can make much more complex decisions.

Like a combination of programs operating in unison.

Yes, clusters of neurons and clusters of gates can do very similar things.

I thought you were more Skeptic Oriented, but you speak about AI as if it is possible.

Of course it's possible. We have a ways to go until we get there, though, both in terms of processor power and programming.

Since you claim knowledge of programming (seems unlikely given your stance), tell me how a computer could ever be programmed to imagine or invent a solution not in its programming. Intelligence should be able to do these things.

===================
Artificial neural network

An artificial neural network is an interconnected group of nodes, akin to the vast network of neurons in a brain. Here, each circular node represents an artificial neuron and an arrow represents a connection from the output of one neuron to the input of another.

In computer science and related fields, artificial neural networks are computational models inspired by animals' central nervous systems (in particular the brain) that are capable of machine learning and pattern recognition. They are usually presented as systems of interconnected "neurons" that can compute values from inputs by feeding information through the network.

For example, in a neural network for handwriting recognition, a set of input neurons may be activated by the pixels of an input image representing a letter or digit. The activations of these neurons are then passed on, weighted and transformed by some function determined by the network's designer, to other neurons, etc., until finally an output neuron is activated that determines which character was read.

Like other machine learning methods, neural networks have been used to solve a wide variety of tasks that are hard to solve using ordinary rule-based programming, including computer vision and speech recognition.
=======================

Here at my company we are working on a similar system. Some details:

http://www.engadget.com/2013/10/11/qualcomm-zeroth-brain-like-computer/

There are only 2 types of computer coding. Comparing 2 things, and directing an instruction.

Google "multiplier" or "associative cache." There are a whole lot of things that computers can do beyond "comparing two things" or "directing an instruction."

Brains are not Binary.

They are. Neurons fire or they don't. From Wikipedia - "The conduction of nerve impulses is an example of an all-or-none response. In other words, if a neuron responds at all, then it must respond completely."

However, by changing when they fire, and how often, and what influences their firing - you can get some very complex behaviors.

We can alter our thoughts midstream with a distraction. How do you distract a program?

You provide a strong stimulus to one of its network inputs and then reward any change in its output. After a little while, it reacts to the distraction.

 

[James R]

This is simply a particular reductionist theory of thinking, consciousness, etc.

There is no working model of how they "combine" to arrive at such a point.
In fact there are even suggestions that drawing an analogy between human consciousness and a computer simply misses the point of consciousness being the unifying element that exists beyond mere yes/no algorithms.

The question is: if you think consciousness is not a product of neurons in the brain, then what is your proposed alternative source?