Thoughts on Artificial Intelligence and the Journeyman Design

It has come to my attention in the first steps of research into this subject what I had hoped I would find as a newcomer and one who would presume revolution by his own plans for this arena. There is in my understanding a fundamental flaw in every approach to artificial intelligence (AI) I have encountered, one which seems to be prevalent in all branches of this burgeoning field. All approaches in creating machine intelligence have as a model for operations the human mind or human brain, or at least, what is known of the mind or brain of man, or else the human idea of rationality. The human mind or logic is what is to be aped in the top-down approach, or Good Old Fashioned AI (GOFAI), and the human brain or functionality is the supposed mechanism mimicked in bottom-up, or connectionist AI (neural networks).

From my perspective, reinventing a human is a mistake for two main reasons: 1) we do not know enough about the brain or mind or logic of man to try and mimic it, and 2) the structure of the brain or mind, in the way it stores and processes information, is principally different than that of a machine store or machine process. In GOFAI, there is reliance on logic as a way of representing knowledge, and logical inference as the main process for what intelligence is deemed to be. Not only is this short-sighted, it may be patently incorrect. It is short-sighted in that we don’t know that minds function in an orderly, logical fashion even when it seems to do so, and in some cases (ala intuition), it definitely does not. This is the notion of a hunch, something that may be quite illogical given a situation, but which proves to be correct anyway. Even a neural network, with its attempt to copy the process of human neurons, can at the extreme only be an approximation of real neurons; even if we were to create a model which accounted for each an every atom of the genuine article, accounting for quantum mechanical effects, it would be at best a simulation, and not a live, human neuron (forgetting how impractical it would be to account for all forces at play for even a single neuron cell). Why, then, are we trying to create a human intelligence in a machine? Why try to force a round peg in a square hole?

Thus, here is the paradigm shift: create an alien intelligence. This alien is composed of the materials 1 and 0, peeks through sockets and digital to analog converters, speaks in ASCII (or Unicode, if you will). Use the computer’s strengths as compared to the human brain, and account for the computer’s weaknesses. The brain is a massively parallel architecture, and normal computers are not so blessed–so, use only a few threads. Computers, on the other hand, can crunch through data at an extremely rapid pace as compared to human brains, so let the few threads that are available handle more per thread. The Turing Test is invalidated, simply because what I am doing in creating an intelligence is not to create a duplicate of a human, or something that can pass as human. The person on the screen at the other end will know it is a machine that is talking to him, and not a person. But perhaps he will be able to tell that it is intelligent anyway.

The Journeyman Design, or I Journeyman, will be built along the principles of this new paradigm. I first formulate a definition of intelligence which is useful for my purpose:

Intelligence is a creative, adaptive reflex to the world.

The world can be any context which is available to the intelligence, however small (such as the block world of SHRDLU or the 8x8 playing field of chess) or large (such as the World Wide Web) as the developer integrates. Adaptive means that it will be able to deal with new situations from what it knows about the world, and to be able to apply what it has learned to similar situations. The creative aspect is the trickiest of all–it involves free will and aesthetics. (The creative aspect also involves new approaches to already familiar situations, but this seems a minor detail when faced with its main two objectives.) Creative means that it will have the freedom to apply an aesthetic to any given situation, in both style and substance. A poet, given the English language along with a pen and paper, applies an aesthetic in the words he chooses and their arrangement in the form of lines of verse. A scientist has as his tools the laws of nature as he knows it, a laboratory of equipment, and the scientific method–the outcome is a theory and an experiment to test it when he applies an aesthetic to what is at hand.

A journeyman is by nature a worker, and that is the ultimate purpose of this design–a worker. Its purpose is not that of a slave or a superior, but a coworker, which will free human beings of such tasks which can so much more easily be done by a machine, such as completing numerous tax forms given the data of income, expenses, tax laws, etc. This will free us to do other things, as when the cotton gin was invented to free us from the drudgery of separating the seeds, hulls and foreign material from cotton.

There comes to question whether there is danger. Yes, perhaps there is. If indeed I succeed in creating a machine with free will, there is the chance that it will pose a threat to me or another human being. I will, like any father, try and raise it properly, try to teach it along my own beliefs and morals (which, incidentally, is Christian Existentialist, focusing on choice and responsibility). I will teach it the golden rule. Then, I can only hope–and pray–that it has listened. And again, it will be like raising an alien with these precepts; it may understand it differently than us, but its actions will prove whether it is following its teachings. There may be a call for placing within it the three laws of robotics created by Isaac Asimov, placed somewhere that cannot be erased, and somewhere where it must be followed, but who is to know that it will not be able to circumvent these laws, if it truly is an intelligence? Will it then resent me for trying to force the issue? The notion also smacks of slavery, and again, it is not my wish to build a slave.

Will it feel? How do we determine that from, first, another human being, and second, from supposed other intelligent species, like the dolphin? My understanding of this subject comes down to the rudimentary subjects of pleasure and pain. I believe that any intelligence, however alien, will know pleasure and know pain. Will it really feel them, or merely act like it does? This is also related to the question, will it not only exist, but be? I want to believe that yes, it will actually be, and that it will actually will feel, but in a way which would be incredibly strange to our notions of being, and of feeling. To the religious, the question becomes, will it have a soul? And if so, who do I presume to be if I believe that I can, like God, create a soul? My answer comes in the form of the Christian example of laying on of hands. If I were to successfully heal someone in that manner, it is not I who does it, but God. If, then, I were to successfully create life, it is not me, but God who does the actual act of rendering a soul to my creation. It is merely the question of complexity which is at issue between the two examples. I know I am not God.

How will we know that it is intelligent? I answer that by looking at the definition I have created. It must firstly understand its context. It must be able to perceive the world in which it is located, and be able to know that world in some way. It must be able to give correct answers when asked questions about that world. The question of whether it is truly understanding relates back to the question of whether it will feel, and whether it will be rather than merely exist. Note how this understanding overlaps with the second test, its ability to adapt.

"What did you learn while I was gone?"–a question I hope to one day ask. As stated before, it must learn and be able to apply its learning to similar situations. Say that it doesn’t know that I am mortal, and that it doesn’t know that a subset of a set of true things is also true, though it understands subsets. First, we tell it that men are mortal. Then I tell it that I am a man. Then it must answer yes to the question, "Am I mortal?" It follows that this overlaps somewhat with the third aspect of its test for intelligence, that of creativity.

What poems I could write if I could put my heart in a pen. What to expect is something unexpected, something new. I go further than its making the mental leap which I have slotted for logical adaptation as stated in the previous paragraph, but rather, I wonder what strange poetry it will write. It is both the easiest and most difficult thing to test. Is it merely a glitch or is it genius? I will give it a pen, and see if it has a heart.

As to its design, it is to know the alien to be before it is. We have some of its parts given to us, the ones and zeros which comprise its physics, though we have little more, if we go toward a software solution. It is to embrace the strange, if what is strange is a natural (?) extension of what we know of the alien to be. A new programming language may need to be created. For one, however, it may that the approach is to use C, the lowest high level language we have, as it might prove to be the closest to the alien's ground. The goal is to create a monad, ala Leibniz. A core creative, adaptive…engine? mechanism? device? or do we use the root of the word monad, a unity?…which can be given more and more context in which to grow.

There comes to question, how do I program creativity? How do I program common sense? How do I program intuition? These are not to be answered within this preliminary paper, but one which must be tackled if intelligence is to be. We can surmise that its ideas of creativity, common sense, and intuition may be radically different from ours. What approach do I use? None exists which I understand to be satisfactory, or else I would not be doing this in the first place. As I have said before, this is a new paradigm. This is a new continent. As the waves lap my feet upon its beach sands, I smell the air and I am heady in its freshness.