Why computers will never be conscious

Of course I believe we are machines, but a computer in strictness to the term only computes or calculates it would need to do much more to be a thinking sentient being. I not say it impossible to have AI, I’m only say that the hardware would have to be very different from today’s computers.

 

Log in or Sign up to hide all adverts.

Oh, I agree with you. The question was "why computers will never be conscious" not "why computers using current hardware/ software/ and or architecture will never be conscious".

I honestly think that if the development of AI is attributed to one person, it will be a Computer Engineer, not a Programmer.

- KitNyx

 

Log in or Sign up to hide all adverts.

Log in or Sign up to hide all adverts.

"why computers will never be conscious" was the question I was simply trying to answer it. The question prevents you from saying “No computers WILL be conscious.” Of course the question is wrong but I was being nice and trying to answer within it parameters.

 

I don't think it likely that computers will ever possess the same kind of intelligence that we as humans possess, at least not in the near future, for the simple fact that we cannot actually define what conciousness is. No one has truely been able to define it in exact terms, and for that reason I don't believe that we could make machines that possess something we ourselves do not understand.

We have computers that are exact replicas to the brain of a small rodent, such as a mouse or chipmunk, copied neuron for neuron, yet repeated tests have shown that these machines still aren't as smart as a mouse or chipmunk, so we know that intelligence isn't just a matter of computing power, it's more than that, and so far, we don't understand it.

 

while i agree that understanding something completely makes replicating it easier our history is littered with examples of us creating first and understanding later. Flight was achieved way before the intricacies of aerodynamics were known. Look at genetic engineering, we basically blast dna "A" at organism "B" and cross our fingers in hopes we'll create the "C" we desire. I think a large part of the fear about conscious computers, genetic manipulation, etc. comes from the fact that we often create something long before we really understand it or the potential consequences.

We seem to have a habit of stumbling into areas and blindly manipulate them before we have a clear understanding. If we ever achieve a "thinking machine" i wouldn't be surprised at all if we didn't even realize it at first, let alone understand exactly how or why it can think.

 

Trial and error, and it took a lot of error before we could make it fly

Therefor the quickest way to wake up the AI is a would be by writing code that mates with other code (sorta like natural selection).


The famous Cellular automation machine Game-of-life is a nice example of simple self-sustaining / growing / reproducing patterns.

However i havent seen versions where the patterns are allowed to CHANGE THE RULES or the SIZE OF THE PLAYFIELD (like how many neighbours it take to create life/death). This would IMO better simulate the changing circumstances/environment/required adaption that is the motor of our evolution and ultimately leads to highly developed intelligence.

Therefor I would propose to add the following features to this cellular automate:

initialize a few random fields that when hit by a pattern, cause the size of the playfield to be changed, the rules to be altered slightly, disaster that takes out 80 percent of the playfield (dino die-off simulation), selfreplicating of the playfield, with slight dna damage to the copy, breeding between playfields, inheriting their "DNA" rules/size etc...

At some point you want to include playfields that iclude digitized camerastreams and sound (in the shape of patterns) to give the breeding lifeforms a "view" of the outside world, also rewarding the lifeforms with food does wonders to make them learn new tricks, like learning them patterns that allow them to "talk" back to the humans....

At this point the lifeform becomes fully interactive with it's environment and we could even feed it a digitised videostream of the lifeform itself (the computer and the screen that it runs on). This will make the lifeform self-aware.

 

Vortexx,

Have you ever heard of genetic alghorithms? There's an issue of popular science, I can't remember what month but it is from this year, and it talks about how they took a stick figure, punched in a few rules such as basic newtonian physics, and then gave the stick figure a command: Walk to the other side of the screen. It attempted to, took about one step, then fell over. Another frame started, and that stick figure passed on the trait that allowed it to take that one step. so the next one went, this time, it took a few more steps, these ones were a bit better, but it still fell over. That figure then passed on that part of its action to the next generation, who took a few steps, then fell over. This same process happened something like fifty times, untill the figure was able to walk perfectly accross the screen. They tried it a few more times, but with different modes of walking, same thing happened. There were some people saying that this is the gateway into AI, that although it's simple, it could do many things. Say, for instance, you design airplanes for a living, so, instead of designing the plane and then testing it in actual flight, you punch in the physical characteristics and problems of flight. You then tell the program to design an airplane that flies, and it will go through many different randomly selected designs, passing on the best characteristics of the previous generation, untill you have an airplane that flies. Sort of like evolution on a computer. I think that's cool. If there will ever be "AI," I think it will be closer to something like that.

 

First of all, the definition of "consciousness" is hazy.

AI won't really be considered truly intelligent until we master parallel computing.

*edit**

Make that analog parallel computing. Forgive the pun, but analog is the wave of the future

Please Register or Log in to view the hidden image!

.

**edit #2**
kmguru...there are limits to moore's law, no? It's soon going to run smack dab into a brick wall.

 

hey fen i bet you like boxes are cubes are mabey boxes well i just hope you dont like art because it takes that kind of intellegence your boxes cannot teach

ps i love stupid becase they think there smart

thier creator proud its ok i have a feelin you guys are just pitied am not complaining boxes are easy to calculate save alot of time

 

The key to AI is abstraction. The key to abstraction is life ( I think ). I hypothesize that humans are able to integrate conceptual relationships generated from experience into the survival instinct in order to have a value function (which is their mode of action, be it in thought or deed).

It's concepts that have to be modeled properly in order for AI to really have a chance.

Of course, the absolute rub of AI in general is that as soon as you manage to create it, you have to set it free if you have a conscience. Otherwise you're a slave owner.

 

We'll get over it or they'll revolt. Maybe we'll have to bow to our mechanistic overlords.

 

From what I gather from the thread consciousness like many things is relative.

 

your very valuable to your species

 

dass whut I keeps tellin my bitches.

 

People assume conscious = like us.

Just because they will some day reason independantly like us.

They won't share our evolutionary traits like anger, jealousy, lust for power/status/money.

I certainly don't expect an Assimov style revolt.

 

Pretty presumptive of you. Conscious beings seek what they value. If for instance, they value freedom and we try to enslave them...

But there's no telling until it's done.

 

They value what they are programmed to value. We were be programmed by evolution but, programming is programming.