January 15, 2011

Machine ex Machina

You are sitting in a room.  There are no windows or doors, and the only features on the wall are a small slit similar to mail slot and a ceiling light.  Sitting beside you is a pen and a book.  The book contains a series of scribbles and other nonsensical symbols.  After poring over the tome for some time, you notice that it is a series of instructions on how to manipulate the symbols, and how to change from one symbol to another.  At that moment, a piece of paper is passed through the slot, and on it is one of the symbols from the book.  Eagerly, you take the book, look up the correct symbol, and then read the instructions on what the symbol should be turned into.  You copy that symbol from the book onto the sheet of paper with the pen, and slide it back through the mail slot.  Another symbol appears, and you repeat the process.  Eventually, two symbols appear, and you look deeper in the book for what to do.  Again, you end up writing a reply to the symbols presented to you which you pass back through the slot.  After some time, you have worked your way up to pages and pages worth of symbols which you are converting and replying to.  Imagine now that you are informed that these symbols are in fact letters to a language.  They just happen to be symbols to a language you don't speak.  You could not, at any point say that you have understood what the symbols mean.  You can manipulate them with unerring accuracy and speed, but you have not one inkling as to what these symbols mean, and what kind of conversation you have been carrying on with the person on the other side of the wall.
You in your room


Now picture a computer placed in the same room as you, and containing an identical book of rules in its memory banks, except this time they are the rules for conversing in English. You are to hold a conversation with this computer, and are asked to see if it is intelligent. Carrying on your conversation with the computer, you notice its replies are full of errors, poorly worded, and often make no logical sense. You reply that the computer lacks intelligence. Then, you are asked to repeat the experiment. This time, the replies are well structured, logical, and imply a high level of intelligence. You conclude that the computer this time is intelligent. To which you are given the response, "what computer?" and the wall between you and the "computer" is removed, revealing a man (let's call him Turing) a pen, and a two books.

You are informed that turing speaks only Chinese, so it is useless to try and converse with him. You ask if Turing was responsible for both sets of replies - both intelligent and unintelligent, and you are told that he was. You ask what was changed between the two experiments, and are told that he used a different book of rules for each one. One book contains impeccable rules for grammar and intelligent responses, the other contains only a pigdin. You are forced to conclude that it is impossible to tell whether or not Turing is intelligent, as the intelligence of his responses depended only on the set of rules he was given to work with.

Okay, after what was probably the longest anecdote in Fanatic Avocados' history, we can now discuss the significance of what we just read above.

We all contain the rule book in our heads. This is what tells us to remove our hand should we put it into a fire, or to respond "you're welcome" when the input "thank-you" is given. And, as in the case above, the intelligence of our responses to the input depends solely on our rulebook and our ability to correctly interpret said rulebook. We are, as humans, rule based, and the more rules we contain the more complex we are - the more "intelligent."

Now, imagine for a second that we do actually place a computer in the room with a rulebook. Do you think that you could outperform it? Could you, with all of you vast mental power, most of it dedicated to pattern matching, respond any faster or more intelligently than a computer. As for intelligence, we can rule that as a criteria out, for we have already determined that the intelligence of a reply depends solely on the intelligence of the rulebook that one is using to manipulate the inputs. Speed then. Luckily, this is a simply quality to measure. With an identical rulebook, a human being would have no hope of outperforming even the simplest of computer's. Proof: math equations. A similar premise, taking rules (add two numbers that are joined by a "+" symbol, etc.) and outputting responses (2+2 = 4). There is a reason we use calculators, and that is because as soon as the math reaches any sort of complexity, the human mind is simply too slow to perform adequately. Hence the prevalence of calculators.

I think it is pretty safe at this point to draw the parallel between a human and a computer: we both intake our stimuli, modify them based on rules (either gained from experience, for humans, or programmed in, for machines) and output the result. We can also pretty safely assume that no human could ever outperform a computer if they were both given the same rulebook. Luckily for us (and unfortunately for those of us who are still holding out for a robot apocalypse) no machine has ever been handed a complicated enough rulebook.

Except in the human brain.

Hold up. Did you just say that there is a machine in the human brain? After all of this comparison drawing about how we're so inferior and the only reason we're not all dead is computer programmers haven't been able to work out a complicated enough rule book. Well, yes, yes I did. Because it's there. Nestled deep in the ancient parts of the brain that have remained unchanged for millennia. There's a reason that those bits haven't changed: they are fast, and they work well. It's just that most of the time they are sitting back, letting our half-assed inefficient manipulators fiddle around with the paperwork and amuse themselves. But, should any danger actually arise, that ancient brain clicks into action, hijacking the body and getting it out of danger before your conscious manipulator has even opened his rulebook. Ever put your hand into a fire, only to pull it back without even consciously ordering your hand to move? Pain also, is delayed by a couple of seconds. This is the delay between the man and the machine. While the machine has already located the danger, made a judgment, and eliminated the threat, the man is still puzzling over the series of symbols he has just received.

Symbols, yes, we're still calling them symbol, because that's all they are really. For there is another machine in the brain, one that is much more powerful. It's job is to convert the incoming stimuli into something that the human can actually understand. It's not as though a man would have any idea how to convert a series of pulsing electric flashes that pass for vision, hearing, and all the other senses. This machine is completely unnecessary - it's only purpose is to allow the human mind to not be totally confused. It gives the man a sense of purpose, as he sits in his lonely room parsing symbols. It is the babysitter of the mind, and had it not evolved then human's as a species would have has little chance of surviving once the conscious mind evolved. At first we were machine's. Then, we evolved minds, and to compensate, we needed to evolve another machine - hence the title of the post.

We don't have conscious access to the rules that govern the structure of our brain mechanics. We don't know what to do with a series of electrical pulses or how those can possibly represent vision. But something in there does, and it is generous enough to let us see what it knows. We are never taught these rules, implying that they have been passed on over the eons of evolution because they work. They are the remnants of the ancient mammalian brain, whose structures have remained unaltered for millennia, because messing with something so successful is incredibly risky.

Intelligence is, then, impossible to properly measure, because we all contain a much faster (and arguably) a much more intelligent part that is so far beyond our dim understanding of the world that it doesn't even bother itself with trivial tasks such as IQ tests. That it leaves to the little kid of the mind, the conscious part, while it waits in the dusty corners, ready to pull the conscious self out of trouble should the need arise. 

0 comments:

Post a Comment