What If Artificial Intelligence Never comes?

by metatron 23 Replies latest jw friends

• 

  hamsterbait

  Until we really understand how the neural network of the brain produces intelligence and consciousness, we will not be able to create technology to do the same.

  At present, efforts are directed mostly to computer programmes that make a machine act as if it were intelligent.

  Artificial intelligence is prolly not a simple matter of how many connections or switches a machine has.

  The first computer, the Analytical Engine was entirely clockwork. If a sufficiently high number of cogs could be made to generate consciousness, it would be so slow you would have died long before the display clicked up "Hi"

  HB
• 

  jgnat

  I hear what you are saying, that perhaps conscousness is the unique creation that sets us apart from the universe.

  Or maybe consciousness is a simple defect in our design.

  There were two paragraphs in your article that caught my attention:

  Consider the low-energy listlessness that accompanies melancholy, the overflowing jump-for-joy sensation that goes with elation, the pounding heart associated with anxiety or fear, the relaxed calm when we are happy, the obvious physical manifestations of excitement--and other examples, from rage to panic to pity to hunger, thirst, tiredness, and other conditions that are equally emotions and bodily states. In all these cases, your mind and body form an integrated whole. No mind that lacked a body like yours could experience these emotions the way you do.

  I've always thought so myself. Even if we DID build it a perfect brain, if we did not give it senses, if we did not give it the instinctual drive to learn and discover, wouldn't it still just be a box? Just as, if we put a human developing brain in a box (heaven forbid) it would be as flat and devoid of character?

  But if we take the route Turing hinted at back in 1950, if we forget about consciousness and concentrate on the process of thought, there's every reason to believe that we can get AI back on track--and that AI can produce powerful software and show us important things about the human mind.
• 

  zagor

  Excuse me, but I think we need a disambiguation here. Artificial Intelligence (AI) and consciousness are two different things. Would you say an ant or a fly have consciousness in sense we people perceive it?
  In fact, how do we even know they have any consciousness at all? Well one way of looking at it is the way how they avoid obstacles which would indicate some sort of reasoning power in their little brains, right? Or how about swallows who can navigate hundreds and even thousands kilometers with no maps. So if we managed to create a machine that is autonomous and able to make similar intelligent decisions within its own domain that would be the evidence of at least artificial intelligence, wouldn’t? I say within its own domain i.e. environment because we expect nothing more from animal either. Say a Tuna is not expected to be comfortable making decisions while flying through the air, right?
  So did we make machines capable of emulating such intelligent behaviours? You better believe it. Would you say a driverless car was exhibiting intelligence if it was able to drive by itself some 500 miles? Or even through rush-hour traffic? http://en.wikipedia.org/wiki/EUREKA_Prometheus_Project

  http://www.innovations-report.de/html/berichte/automotive/researchers_develop_intelligent_driverless_car_81223.html

  In fact, today we are going beyond creating mere robots and are well on our way to create those that make the decisions the same way many animals and insects do. A very good book, I strongly recommend you to read is Biorobotics from MIT Press http://mitpress.mit.edu/catalog/item/default.asp?ttype=2&tid=8578

  Or how about robots that exhibit social behaviors? Would we consider it to have AI if it was able to read our face and by extension our emotions or able to show similar emotions to the things we do around them? http://robotic.media.mit.edu/projects.html

  Heck, we have entirely new fields of Biomechatronics and Cybernetics that have a vision of nothing less but creating a cyborg http://biomech.media.mit.edu/

  As far as consciousness is concerned, heck we don’t even know what is that in humans. We can only assess it externally. If someone intelligently can answer to a question we assume he got some cogs in his head if he can’t we assume few are missing.
  Likewise, how would we even know if there was conscious thought processing happening in any of our artificially created machines?? Well strangely enough the same way we assess that in humans. We can’t open up their skulls and see some ethereal substance moving that we could say. “Aha there are thoughts happening”. Lol.

  Hence, like in humans we have to accept that those machines will make good and bad decisions based on the amount of power their brain can process. And you know what?! That is exactly what we are seeing in many of advanced systems.

  But first we need to define what consciousness really is both in humans (taking into consideration varying amounts from “normal humans” to savants) and between different animals and insects.
  Before we are able to precisely define it any assessment and comparison is meaningless. Do I believe machines are conscious in a way we are, no I don’t. BUT, and that is a big but, we have to accept that there are varying degrees of consciousness. Because I would assume that animals and humans are very different how they see the world around them. So the answer is : of course we haven’t recreated conciousness in the way humans perceive it but we are moving fast. And to go back to the same point again. How would we even know how to assess the consciousness if we saw one?
• 

  zack

  I think scientific progress is wonderful, but it misses the point many times. For instance, any human being with an IQ of 60 or less is considered an idiot. He needs help. He makes poor decisions. He is incapable of problem solving beyond his conditioned and simplistic learning. Yet, he is fully a human being and fully concious, and yet we lable him as something less than what we ourselves are.

  If a machine could operate, think, interact, and feel like any human being with such a low IQ, we would be astounded by this "miracle" of science, yet we tend to discard the very same human being as something less, a burden on us, a blight on our kind.

  I love science. But it has no soul, man.

  As for smart machines, they really are all around us. If they expected to mimic a human being, then I do not beleive they ever will. A human making another human is possible---- you just got to like the infant stage of development.

  Good thread. I always enjoy your subjects.
• 

  IP_SEC

  AI & consciousness are two different things.

  No not necessarily. If a machine can simulate a human intelligence there will be no way of knowing if it really is conscious or just simulation. For all we know, our brain just simulates consciousness.

  It is a common myth that AI failed. We already have very specific AIs that perform better than HUMAN at the task it was designed for. More general AIs are soon to come. When the neural net is mapped. Simulation of human consciousness will not be far behind.
• 

  zagor

  If a machine can simulate a human intelligence there will be no way of knowing if it really is conscious or just simulation. For all we know, our brain just simulates consciousness.

  ... which is exactly what I said :)

  We already have very specific AIs that perform better than HUMAN at the task it was designed for. More general AIs are soon to come. When the neural net is mapped. Simulation of human consciousness will not be far behind.

  Not sure what you are refering to when you said "when neural net is mapped" I work wiht AI and neural networks every day. Nerual networks we use today whether you talk about Backpropagation, Fuzzy ARTMAP, Stochastic Neural Network, etc are only broadly based on biological neural network. For instance backpropagation is probably the most natural (though it takes longest time to train) it uses multiple inputs that are fed into individual neurons in hidden layer(s) sum obtained is then fed into squashing function that can be simple step fuction (i.e. binary 1, 0), sigmoind or even something else. Such signal is them passed all the way to the output upon which it is calculated against desired output and backpropagated for each individual nodes and weights recalibrate. In my own experience even a simple network can take between 60.000 to 120.000 cycles to train. Even with computer speed that is a helluva long time. Here we use Stuttgart Neural Network Simulator which is probably one of the best open source software and it still takes long time (you can go hare download it and try for yourself http://www.lix.polytechnique.fr/~liberti/public/computing/neural/snns/) I'm writing all this to tell you that this is not exactly how biological neurons work (i.e. only to a certain extent)so any mapping of brain tissue won't do the trick, unless you want elaborate what you meant. But you are absolutely right even if machine started thinking we would in most likelihood be blissfully unaware of it.
• 

  zagor

  as a side note:

  You pretty much said it yourself if consciousness appeared we would have no way of knowing because there would likely be no outside manifestation, anything that would happen would happen inside of machine's brain, But there are manifestations of AI in terms of intelligent decision making which is something we can see based on what machine (robot) does. So yes, based on that those two terms are different.

  I think problem arises because we look at those intelligent machines in terms of human intelligence which for the most part doesn't separate consciousness and intelligence
• 

  Paralipomenon

  If we were unable to create real artificial intelligence then I would agree, but mankind is still youthful in it's efforts.
  
  We operate off a binary system, everyone knows it, 0 or 1. On or Off.
  
  This is very similar to the told telegraph systems. In order to send a sentence you were required to break down the word into a series of dashes and dots. Great for sending short texts, but ineffective for holding conversations. It wasn't until the telephone that used waves and harmonics that greatly sped up the process.
  
  Our current computers are still figurative telegraphs. The processor is a bottleneck. Sure you can add more processors so you can handle an increased volume of 0 or 1's, but this creates a huge bottleneck when you want to attempt something such as real time consciousness. The more machines you add, the greater the problem of having them communicate with each other.
  
  Several proposals have been worked on, two of my favorites deal with fiber optics. Right now they are being used as a standard on/off blinks of light, but the realization is there, that that blink of light can be any shade of color. Rather than On/Off, the same pulse can now be one of millions of shades of color.
  
  But it would be expensive and the old binary system is still suitable. We haven't tapped out our potential on that.
  
  This would provide an absolutely exponential increase in processing power, but would still remain a sequential system.
  
  One command = one pulse.
  
  There needs to be a way to send concurrent commands in a single pulse and I think the next step after optic computers will be chemical/biological.
  
  Back to the original point. If we create intelligence using our brains as models and end up using the same process, can we really refer to the intelligence as artificial? Or have we just become gods?
• 

  IP_SEC

  zag, by mapping the NN I meant the human brain.

  We operate off a binary system, everyone knows it, 0 or 1. On or Off.

  By "We" I assume you mean the human brain? The human brain is a hybrid binary/analog computer.
• 

  Paralipomenon

  By "We" I assume you mean the human brain?

  Not really, "We" as in mankind. Our current technology is based on binary.