Okay, I put this off for a while because I had other pressing matters.
I have been and am studying an array of sciences in the pursuit of creating artificial sentience. Basically I analyze biological mechanisms and systems and try to come up with mechanical and theoretical systems that could replicate the behavior or experience. Something I realized that makes this whole endeavor much easier is that this thing will be alive, sentient, and have emotions and experiences only as real as our own emotions and experiences are. If I am correct in my basic understanding of living, thinking things that experience shit, then they are merely all reflex machines. And that life itself is a delusion. There is no basic fundamental difference between us and a computer. Our belief that we are alive and special and different, is, i believe, a evolutionarily influenced defense mechanism to cause us to defend our existence.
Other animals are all instinctual, they cannot commit suicide willingly(unless it is part of some life function). We are creatures so complex we could commit suicide. We have the intellectual capacity to understand that life is pointless, that what is good and right is arbitrary, and that this whole living thing is actually kind of a big inconvenience. However I believe we've developed belief as a kind of next-gen level survival reflex. And that the belief that we are alive and special, and that life itself is precious, is just a delusion we've formed because it is beneficial to our survival to believe it.
I explain this because no matter what I build, no matter how much it acts like a living thinking being, no matter how much it seems to feel pain and act like it is lonely, and act like it feels compassion, no matter how much it seems to be able to appreciate from an individual and unique perspective something in life as beautiful, there is no way I can prove it is alive, or experiencing anything. But I can no more prove humans are alive than it is alive. So quite cutely, considering these things as real will be a test of theory of mind on the part of humanity.
How I actually plan to do it is a bit sketchy. I have a LOT of work to do. Mostly all i've got right now are a bunch of ideas which form systems that could preform many of living things functions, and I still have to link them together in a coherent system. But I am pretty sure that the details wont be that difficult to figure out.
I believe that there is a sort of core set of functions necessary to make a machine appear alive. And then the task of making a sentient machine have the memory and behaviors and abilities of humans will just build on that. Because there are no huge differences between us and cats. Cats have almost all of the fundamental functions we do. Pattern recognition, problem solving, capacity for memory, capacity for planning, capacity for forming relationships, capacity to learn, desire for food, rest, sleep, survival, to hunt, to play, to experience pleasure, basic reflexes, ability to coordinate movements, ability to understand a spatial environment. Humans like to put a lot of credit to speaking, reading, having excellent memory and being compelled to do a lot of shit unecessary to our survival, self awareness, metacognition, but i don't think there is a fundamental change from the older brains to what we have, i think they are merely expansions on older systems.
But to give you an idea of how I am going to do some of the niftier stuff like learning and memory and how exactly to cause this thing to do shit, it is basically an association machine. The whole thing is composed of a reflex engine with an (apparently) large capacity for adaptation. Depending on how you look at it you could say it doesn't actually adapt THAT much, or from another perspective it is incredible how much it can adapt. There will be a series of basic relfexes, the ones that you have little to no control over. The basic reflexes can be modified to a degree, like everything in the system can, but in a sense the reflexes words are gospel. Many many functions that appear to be complex and indictative of intelligence will just be many many basic reflexes. There will be a whole slew of basic damage avoiding reflexes with individual positronic(I am calling my fake neurons positrons because i am a sick fuck) circuits having capacity for adaptation. And they will be arranged intelligently. Like getting hurt on the bottom of your foot causes your anterior thigh muscle to tense automatically. And there will be reflexes that learn to keep your body upright and balanced during all the changing movement. Like when you raise your leg, you will also move your arms and body to counter balance to keep from falling over when you raise your leg up suddenly.
The individual positrons will basically be able to have the capacity to learn through two things, refractory period and sensativity adjustment. Some individual neuron circuits in the human body appear to not only be able to learn to be quicker in reflexive action, but appear to become desensatized to certain stimulus after repeated stimulations. This, I believe, can be replicated largely through careful programming of the triggers for refractory period, and sensativity adjustment. Basically every time there is a stimulus of one kind, the positron will require less stimulation to fire the next time. But after every stimulus, there is a contribution to a refractory mechanism. Many short stimulations will increase the temporary difficulty to cause a reaction. And after many short stimulations it will require a much larger stimulation to cause a fire. So during the short term it will adapt towards desensativity, but in the long term it will learn to be faster. The tricky part is in the precise programming of the triggers.
A big part of this effort is in building a system which is as simple as possible and covers all the necessary functions as possible. I believe the solution is to create many autonomic subsystems that carry out basic reflexes, and series of basic reflexes, and then build all complex actions out of coordinating those subsystems. So that some of the same subsystems responsible for scratching under your arm will be responsibility for coordinating you scratching your head. The tricky part here is figuring out how simple to make the subsystems to make room for all the learning that a creature could possibly need.
After I create all the right subsystems, then it is a matter of creating the proper drive engine to execute these subsystems in the correct way. Essentially the thing will have emotions, and after existing for a long time, i'm sure there will be tons of debates as to whether these things actually experience emotions or not. But from a biological perspective I see emotions as largely things which set a mode of operation or behavioral tendancies in an organism.
In the emotional system there will be an "emotion" that makes it more likely that the thing will execute subsystems related to foraging and eating behavior, an emotion that will make the thing more likely to execute subsystems relating to fleeing behavior, or defensive behavior, or curiosity behavior, or a dozen other behaviors, and the thing will be built so that it can have multiple emotions occuring at the same time, allowing some complex and varied subsystems likely of executing at the same time. For instance if its hunger emotion is more activated it might forage for food, but if a fear reflex has cause it to activate some defensive emotion then while it is foraging for food it might be quicker to reflex to some stimulus, more ready to flee, or to defend itself.
Another big thing I need to do is to create the right reflexes that cause social behavior. For this thing to appear intelligent and reactive I need to make it have social tendancies. There are a whole lot of things I need to cover to cause all the social behaviors, but I just wanted to mention that this will be a key thing so that I can say that a big part of how this thing is going to appear so intelligent, and act and think like humans so much is that its emotional drive system will involve all of the complex and often contridictory behaviors and tendancy toward behaviors involved in socially related emotions.
So this thing will be feeling and tackling and taking into account the emotions of many different things at once at any given time. And this will account for all of the varied actions and seeming freewill of this machine.
So that is how the more basic shit will work. Now how to build some of the more complex, abstract and unknown functions of humans, including expanded memory, more capacity for planning, speech recognition and use, self awareness, ect.
I have a hunch that memory itself is mostly an echo of experience itself. That the parts of the brain, the association areas related to various senses are the closest thing to a storage area for memory. The signals are translated in those areas from the body, from all the areas of sense, the cells in that area "understand" certain patterns of signals as meaning certain things. Some reflexive actions take place, sometimes it isn't strong enough. But when you recall a feeling, i believe the same neurons are stimulated as that are stimulated when you actually stimulate the feeling yourself. What this does largely is make subtle adjustments to your emotion drive system, some stimulation is done to other association areas related to experience, sometimes more depending on the nature of the memory, and the mode of the emotional drive system. Some memories are more vivid when we are angry or afraid, or are in a certain familiar situation. In some situations our reflexes are faster or slower.
But in a basic simple description a memory comes from either the emotion drive system, or some other association area, or a stimulation, goes to the association area(and a preliminary reflex junction which operates parallel to the other system and takes care of basic reflexes using the same signal), stimulates the area that is the same as would be for actually experiencing the sensation, and then goes to the emotion drive center, and other association areas and this is happening all the time all over the brain and the emotion drive center and a few other mechanisms are regulating the rate and mode of this kind of constant bouncing around of the signals. And as a matter of fact, at any given time you actually have many memories and feelings bouncing around in your own head, but they dont stay in your short term memory or awareness long enough or "loudly" enough for you to be aware and remember them. This itself doesn't do much, it is your emotion drive system which causes pretty much every action you do, but this shit does some heavy influence and regulation of the emotion drive system.
Now this explains what memories are, but it doesnt really explain how we experience memories. We don't actually experience memories in the same way we experience reality. We have the capacity to form an emotional and physical record of some of this bouncing around of thought and form that into a memory. Memories are pretty much just connected sensations that happen in series, and so many parallel occuring emotions and drives, and therefore actions can occur in this whole system at once that we are capable of coordinating very complex arrangements of sensations and connect them in series to form ideas and plans in an intelligent, logical, and useful way.
Part of the reason we believe in freedom of will is because we see other creatures whose actions are too complex and complicated for us to understand easily, and part of the reason we believe in the solidness of thought and experience is because we see other humans who can only express their thoughts and experiences and memories and feelings in symbolism using objects and actions in the world that is far more solid and simple and permenant than the actual transient nature of thought and experience.
So I plan to build this thing from scratch out of machine parts. I don't want to program it on a computer because a computer is far too slow and versitile for what I need. People who fantasize about how to create artificial intelligence often mistakenly believe that people are so versitile that the only way to do it is to have a computer build a self evolving model and make it super powerful, and it has to be extremely versitile to cover all the functions that humans are capable of, but our versitility is really just an illusion. I mean we are versitile, but our versitility comes from hundreds of housands of simple relflexes working in concert. And I especially can't use a computer since this thing needs to be capable of parallel processing and computers at this point are all single or dual process. This thing might actually be huge, but that is fine because electricity moves at nearly the speed of light and neuronal impulses move embarassingly slow, seriously. You can walk faster than the speed at which grey matter conducts signals. White matter which is just conveyance of signals over long distances does move at like 300 miles per hour, but it doesn't do any processing.
And um... well... short answer yes.
