Saturday, April 21, 2012

Us from Computers (part 8) Gorillas and Siri

Last updated April, 21, 2012 at 4:45 PM to include a few forgotten points.


[Jump to part 1, 2, 3, 4, 5, 6, 7, 8]    



     Let us first look at Turing’s machines.  He was saying that any reasonably complex machine, could with the right set of rules, be mistaken for a human mind during a five minute game.  The rules to the game are simple, two people and one machine are set down in such a way so that one of the people cannot see the computer or the other person.  That person is the interrogator that must, through questioning, determine which is the human and which is the computer.

     To Turing, being mistaken for a human 70% of the time was the same as being functionally equivalent to a human, or at least reasonable so to be able to declare that machine/program combo to be a thinking machine.  One point of interest is that Turing was only talking about digital computers, which has been a point of contention for some time for philosophers examining his arguments.  Despite the semantics of it, if his test of intelligence was a good one, then it should be able to be applied to anything we think might be intelligent. 

     We should be able to run his game with dolphins, elephants, chimpanzees, gorillas, and even unknown unknowns like aliens.  Of course, the tests would have to modified for unique biology or engineering limitations.  If a computer sounded like a Dalek, or the species being interrogated lacked human-like vocal cords, a human couldn’t help but know off the bat they weren’t talking to a human.

       Nonetheless, a properly calibrated and administered test should be able to at least give us reason to think that the object of the test was somehow thinking, but is the process of thinking enough?  Going back to Warren’s list of personhood, we can see that passing the Turing test doesn’t actually meet any of the characteristics of personhood.  It might give us some reason to pause the next time we feel like throwing our “smart” phone across the room.  If the engineer gave the phone a way to detect physical damage to itself, it would meet number one, with a little more tinkering, Siri might reach number four, and it can already zip through any number of logic puzzles that stump humans on a regular basis.  Think about it, are we enslaving cell phones?

     Again, not really a new problem.  Let’s take the case of Koko, the talking gorilla.  Koko clearly has the ability to feel pain, the ability to solve a variety of puzzles, a wide range of non-genetic or externally controlled behaviors, a vocabulary of 1,000 human words in American Sign Language, and strong indications of a self-awareness.  Does that make Koko a non-human person? 

     If we are going to claim that we are not biased, then we have to acknowledge the possibility that both the “really smart” phone and Koko could be persons in their own right.  What legal rights does that morally obligate us to impart upon non-human persons?  We recognize the legal personhood of corporations now, so why is it so hard for us to see that things already on earth might be due the same consideration?  Again, a post for another day.

     However, merely passing a Turing test, or being a Turing machine, does not grant us the same level of personhood.  It only gives us some comparison between machine and human conversational intelligence.  Why did Putnam think that everything we call a mind is just a Turing machine?  It seemed like a good analogy of how a deterministic mind might work.

1) If a complex enough device could run a sufficiently complex algorithm, then the device is thinking, or showing signs of intelligence.
2) Formalized Logic, the best way to model human intelligence, operates on a set of rules.
3) A complex device can run those same rules and produce the same results as a human mind would produce.
4) So, from premises one, two, and three, it is conceivable to soon see thinking devices that show human-type intelligence.
     Putnam then generalized that argument.
5) Human minds are reducible to Turing machines.

     However, with all analogies, they can only go so far.  As I see it, it breaks down at point two.  Formal Logic is still plagued by inconstancies with human intellect.  It, in no way, captures the full scope of a human mind, which allows us to be rations, emotional, creative, intuitive, and at times completely unpredictable.  Even if we had the optimum computer system, equally complex as a human brain, using the greatest program ever written, the program would be based on a still flawed logic system. 


     An example of what I mean: Gödel sentence G, “this statement is not provable”.  If a mechanical system of logic attempted to prove the statement true, it makes the statement false, but if it tries to prove it false then the statement is true.  Outside formalized logic, we can see that it is a semantically true statement.  We have an intuition about logic that none of our systems have ever completely captured.  Computers are limited by their programs, which a based on our formal logic systems.

     So, what separates us from computers?  Take the best logic system out there, find a way to make it into a computer program and run it on the optimum supercomputer, and then somehow scoop out a mind from one of our clay bodies (if that even makes sense).  You don’t have a human mind, but like Dr. Frankenstein, you have a flawed copy; one that lacks some significant portions of what allowed us to write the logic system in the first place, the ability to disregard formalized rules.

     We make these rules to try and capture how we think, but no matter how far we progress, there is some kind of flaw in our system, at least at the present time.  Usually, it is that our minds can reason past something that our models of our minds cannot.  These models are nothing more than tools we use, and they do evolve, but like the way we use computers, we evolve with our tools.  Our culture changes, our mindsets adapt, the scope of our influence expands, and the reach of our intellect goes beyond what it was before.

     Using the tools we already have generally means following rules.  Making new tools, or thinking about how the tool could do something different, that is an act of true creativity.  It is an act that defies the rules.  It is an act humans excel in, but when we try to express how the process was done, our words normally fail us.  The tools of language, culture, science, and art always seem inadequate to capture the moment when something new first takes form in imagination.  "Rules and models destroy genius and art." William Hazlitt.

     If we didn’t have the ability to ignore the rules, we’d never have broken away from divine command theory.  Our science would have never progressed past Aristotle.  Our artwork would never have moved beyond formalism.  Our literature would never have allowed Shelley to write The Modern Prometheus.  Take that away from us, and we cease to be human.



[Jump to part 1, 2, 3, 4, 5, 6, 7, 8]    

No comments:

Post a Comment