AI, once more
"Here's some inspiration, go put it to good use as I have seen you can do!
"This being the rationale of the matter, why should we limit our
conception of the Divine ideal of ourselves? Why should we say, "I am
too mean a creature ever to reflect so glorious an image"—or "God
never intended such a limitless ideal to be reproduced in human
beings." In saying such things we expose our ignorance of the whole
Law of the Creative Process [our ability to creatively change the
world]. [...] The creative process in us is that we become the
individual reflection of what we realize God to be relatively to
ourselves, and therefore if we realize the Divine Spirit as the
INFINITE potential of all that can constitute a perfected human being,
this conception must, by the Law of the Creative Process, gradually
build up a corresponding image in our mind, which in turn will act
upon our external conditions."As we sharpen our definition of the kinds of beings we want to be, it
only makes it more possible. From birth we have been told ways to be.
As we age and develop we will slowly tear away from those dispositions
and lean in the right direction. This is one of the reason so many
people have difficulty. They have such poor conceptions of humanity,
morality, and virtue so deeply rooted that the flowering moments of
their lives will be scarce and short lived. I may not have that many
grand flowers in my garden, like some brilliant inventors,
businessmen, and other outstanding minds, but I will be sure to have
it well tended with all the prettiest flowers I can till. "
That's an interesting excerpt, and definitely inspiring to some degree. Of course, I'll stick this to the side of AI as you may have expected I would.
AI, as of now, of course lacks creativity. That's probably an issue I tried to explain in my previous email, but failed to come up with the right word. Creativity is an interesting word: if we break it apart, it basically comes down to "being able to create", being "creative", "create-ive". The ability to create is something that humanity should be particularly proud of, because not many other things have it in comparison. AI, for example. So, that proves the quote you provided.
How would we let AI have an ability to create? How can we manifest it to be able to multiply - something that Max said when we were at board and brew was that some of the reasons man wants to learn is because of our eternal death sentence and our ability to multiply. So, really, anything we create can be seen as our child. Even our feelings support that theory; when you build a chair, you care for it. You would feel bad if some dipshit came and broke it. So, creativity can be seen as multiplying.
So, once again, how can we let an AI manifest its own code? Really, that's what lets humans be humane - we create new algorithms every second, in the back of our heads; our creativity is what makes us human. If we give AI a C++ book and a compiler, it can really be humane! A really slow human, due to its hardware limitations, yet able to adjust to its habitat. Slowly.
AI writing itself, adding to itself is something I'll have to think about. Bleh.
ai, again
In order for a human to really learn, it must make choices. Choices are a big step in AI, I feel, as it must make bad ones as well as good ones. It must somehow have a feeling for what's right and wrong. We must give AI an illusion of right and wrong, which comes down to really, giving it feelings. Can we go around this issue? Can we ignore feelings and give it an algorithm instead?
Can we merely create the desire to increment some variable? A happy variable? What would it get out of it? A happy feeling?
What if for every learning experience, a baby got not a feeling of accomplishment, but some upgrade? If we offer a computer a memory upgrade, can we persuade it into doing what we want?
I guess this is the biggest issue: letting it run on its own, giving it enough algorithms so it is able to create its own. We must let it chose, not make an algorithm for choice.
Another issue is ideas. Have you read Plato's Allegory of the cave? In it, he shows a distinction between forms and copies. Eg, a form would be anidea of justice, while a copy is the way we try to incorporate it in our world. A form is math. A form is truth. Math is also a form. Forms are perfect, copies, shadows are not. A copy is merely a shadow of the form, a shadow of the original idea. It's impossible to know the form, the idea, it's impossible to tell it to someone, however it exists.
Question is, does AI know what justice is? Can we give it forms to conceive of, without having to explain it to it through use of language?
ai
We can safely distinguish a few low-level mechanisms of humanity: first, as we established, desire to explore, learn, find out, discover.
So, 1. AI must have desire to learn, know, explore and discover
Secondly, selfishness comes into play through evolution: in fact, I believe it originally wasn't there however it got implanted into our mindset through evolving. If me and you are sitting in a cave and a third guy just brought a rabbit, and we haven't eaten in 3 days, we'd fight to death, as we know no better. So, selfishness can actually be ignored throughout the development of AI, as it's a somewhat bad quality that can lead to severe issues.
On the other hand, helping others and practicing things for the good of a given society can be viewed as selfishness. If I drive on the freeway and all of a sudden I see a guy walking, I'll pick him up. Maybe not completely because I want to help, but maybe because of some chemicals released into our bodies when we help someone. This may be the anti-selfishness system made by evolution, once we solved the bunny problem by creating a knife (now the bunny can be cut up into three pieces).
2. AI must get 'good points' for helping society, and 'bad points' or bad feelings that dwell, for being selfish.
Thirdly, we may consider desire for connection as a fundamental human quality. Note, quality, not idea that we run on. So, this may be a human descriptor versus a low-level mechanism. However, if it's the former, we may want to add it to the first part instead of letting it be a separate one.
Then again, we may instead think of not desire but the ability of communication. Since we are all humans, we all can communicate, in one way or another. If one can communicate, one will communicate, as our brain obviously has some requirement for that.
3. AI must have a desire to communicate with others, be it humans, other AI or animals and plants.
Unknown. What drives us not to give up is the mere unknown that's on the other side (at least that's the case for some people). I know that if I don't just off the roof, I will for sure just be, and being is something I've experienced before, something that beats the unknown just because of its certainty.
At the same time, if we choose to live, then we must also be. One can't just be, one has to do something. That's how we work.
So, 4. AI must be literally afraid of the unknown, aka death.
This brings us to desire to find out what this is. We consider scenarios: matrix, we're trapped in a virtual reality. Some others. This can also be what leads to the first rule: desire to explore. The reason why we want to explore is to find out what this world is and the main question, why.
Are there more fundamental instructions that humans follow without knowing it? Must be. We must analyze our every move, every thought, to find where it came from, why do I want this, and is it relevant.
A feature that seems to exist in humans is that we basically learn in levels. First, we learn physics of the world; well, our body does. We decide that it may be a bad idea to crawl off edges. We figure out how to walk. We balance. And then, that level ends. No more of that learning, those fundamentals are set for life. Level 2: listening and talking. Our desire to communicate & explore leads us to attempt and understand what our parents are saying, and try to repeat. We learn how to communicate, and communicate we do. Level 3: We learn to remember. We now actually form memories of what happens from our point of view. If we learned how to remember before anything else, you'd remember previous levels of your learning, which may (just a thought) give us a hint to what the world is, or just about anything else.
Possibly, it would give us access to parts of our brain that were used to learn previous levels. It may give us the ability to control the rules that were put upon us in the first place.
And this is where the train of thought lost me. I was getting to that AI should not only have two levels of memory (HDD vs RAM), but more! The first stage, learning physics, learning what does what, would be in one level of RO (read-only) HDD space. Then, when memory kicks in, it should be within another level of RW HDD space. And lastly, active learning, the last level that is always learning throughout your whole life, should be a completely separate RW batch. RAM may only be used for active processes.
Keep in mind, this may be changed. We can even have RAM for all the batches (to greatly improve speed), that is regularly backed up to hard drives. Speed shouldn't be a large issue to worry about.
Yeah.
Studying for finals
http://www.devastatingexplosions.com/
Thanks, Sam. So much for all those A's I had.
Success
You know you've achieved success when your computer freezes for 5 minutes while your 8-page long code is trying to compile.
And then you find out it's because you deleted a bracket somewhere in the middle while commenting.
Going green
I punched a tree the other day. Didn't feel a thing, just that my hand went sore for a few minutes.
What does that tell us as human beings?
Women
Either you have a little bit of an asshole deep inside of you, or you're potentially someone's bitch.
Coffeebean parking lot thinks.
Python on iTouch/iPhone
Found a great little guide for install python on a jailbroken iTouch or iPhone: http://gentechblog.wordpress.com/2009/10/07/how-to-python-on-the-iphoneipod-touch/. Even though this writing is immensely old, it worked! Installed Python version 2.5.1 on my G1 iTouch. Now I can happy import antigravity.. not. That didn't exactly work.
Back to class.
Calendarium
It’s raining tags
Me writings
Spammies
Asteroids
