Artificial Intelligence--What issues do we need to currently focus on?

Artificial Intelligence is all about theory of computers, and programming. I write the programs, and the question in the OP is "what issues do we need to focus on?" So if an actual programmer is not allowed to reply, what is the point of the thread?

 

Log in or Sign up to hide all adverts.

We all have two centers of consciousness, but not many are conscious enough to differentiate both. A good primer book is, The Undiscovered Self by Carl Jung. It is one of the least developed frontiers.

 

Log in or Sign up to hide all adverts.

I have a new centre of consciousness. It teaches me things in my sleep. It's strange because it talks to me like a third person perspective, and I answer it with questions. Very odd, and somehow it comes up with stuff that I did not know. It taught me about paradox a couple of weeks ago, and came up with examples, and even let me examine my internal brain structure whilst it gave me examples so that I could check how my brain was reacting to the examples.

 

Log in or Sign up to hide all adverts.

When we walk from here to there, we don't have to think about the details of walking since the data processing is being done at an unconscious level. All you need to do is use a thought based command line like, "walk over there", to get the CPU to run the walk program. The conscious mind only assists with the steering.

If we used the conscious mind to exclusively think through all the details of walking, as you walk, the movement becomes much less fluid. The ego can't keep up with the level of detail and the processing speed needed to be fluid. This is one way to see there are two center at work usually at the same time. The CPU is not talking to you, as you walk, but it works quietly in the background.

The sports star who make the phenomenal play at the buzzer, is all about the unconscious CPU and lightning speed calculations. The ego will get the credit for this unconsciousness, but may not be able to reproduce this with the conscious mind.

The question becomes, is the CPU restricted to just walking and phenomenal shots at the buzzer, or can the CPU also do this with other tasks? Can we use command lines to think, "data process from here to the unknown there", like walking, with the CPU doing the data crunching so it is smooth and not disjointed?

If you look at science, doing research and writing a paper for publication essentially creates baby food so those whom can't get it, initially, can follow. The initial creation, even if correct, is too fast for the conscious mind and causes a panic. It has to be slowed down into baby food for the ego. Full strength CPU output causes the ego to choke like a child eating a steak. You need to cut it into baby size bites and food process it, o it can be swallowed without chewing.

This translation process is an important part of dealing with the CPU, since its data density is very dense. You need to slow it down or decompress the fluid walking into the conscious robot walking, so the conscious mind does not choke. The baby food approach of science is one way and allows the ego to keep up. The image of the unconscious mind talking to someone, in the third person, is not the inner self and the CPU, but one of its firmware acting as a step down. This often will contain data loss compared to direct CPU data. But it may be easier to turn into baby food.

One of the reasons the CPU is unconscious is because of the way the CPU processes data. It makes use of 3-D memory processing. Logic or cause and effect is 2-D memory, with cause and effect the (x,y) axis. The 3-D memory is (x,y,z) cause, effect, cause as well as effect, cause and effect. The second is easier to see. The idea of effect appears, which is a hunch theory. You need to develop the cause and effect logic, to get back to original effect, which got the research ball rolling in the first place.

The 3-D memory of the CPU can be understood as a ball composed of a large number of 2-D rational planes, all on a common (0,0, 0) center, which combined fill in the 3-D volume. If we flex the entire 3-D ball, a large number of logic planes can twitch at the same time, sometimes out of sequence with respect to the 2-D cause and effect on each plane, but nevertheless consistent with 3-D. You need to make one or more of the new logic planes conscious, so the ego can robot walk. Baby food generation helps other see the new plane.

 

short circut

the consequence of

 

A 3-D memory is composed of a large number of 2-D rational planes (cause=x, effect=y), all with a common center (0,0,0) for 3-D and (0,0) for each plane.

For example, the common center might be the best way to stimulate the economy. There is a common goal for all of us, globally or (0,0,0), with a large number of logic planes (different opinions for doing this). The sum of all these opinions is the 3-D ball, which is greater than the sum of its parts; each contains truth but none the whole truth. If we flex the 3-D ball, we can distort various logic planes pulling them into other logic planes. This helps to add the truth within other points of view, into our logic plane. This then redefines the plane.

The brain uses another affect to power the 3-D memory, which can be explained in terms of a 4-D affect; time element. The brain can project to the future, play back how it got there and then adjust the beginning and do it again.

As an example, going into adolescence creates global changes within the body, which last for so many years. The mind is also reacting to the changes and the experiences which are created, allowing various logic planes of 3-D to filled in.

The 4-D factor time projects to the future ( from 13 to 18 years old) using the evolving 3-D memory to alter the body and mind in an integrated way; 3-D. This may not seem fully rational all the time, but it is logical in 3-D. It work perfectly all the time, except when the ego reinforces a 2-D plane or two and does not grow organically in 3-D. We do have will power.

 

I'm working on a Neural Network right now. My first stage is to build a 3D model of a robot, which is half finished. It has realistic joints so that it can be copied to a physical robot. After the robot is built I am going to train the limbs, and senses. An interesting thing to train is the eyes. When my robot looks at a 3D object I have copied a baby's sight, keeping it at short distances. Say 2ft. This will minimise the amount of information that the robot has to gather at the first stage. Baby's often hold onto something like a rattle, and this is an ideal way to hold an object, and rotate it in 3D. Colours are not important, because we recognise B/W photos. So I am concentrating on recreating the image recognition based on human perception. Once the robot can see, I need to work on the hands, and grip. But my main target is to write a neural network based on the physics of my theory, which are slightly different to the physics of science. So when a standard neural network activates its loop code, I will activate a slightly different loop code which includes a different way of using time, and a slightly different way of storing information. The great thing about computers is that I can make a very expensive robot for free using 3D models. I think that the hands that I have designed would cost a fortune just on their own.

 

Pincho, you strike me as someone who hasn't ever done any of those things because each in their own right is an enormous area of research in AI/robotics/data processing, with billions dollars spent on it each year. Something as simple as object recognition is quite difficult if you don't know apriori what objects you'll be seeing. Picking out circles or lines in an otherwise static filled image is simple but something as 'basic' as taking a single picture and then identifying interesting structures or objects is not trivial. You can't just train a neural net to do it. They have short comings, not least because it's all too easy to end up training them to identify your training data, rather than general classes of things. Making a neural net to something basic is itself very basic. It's literally homework assignments for undergrads. The little 'game' you have a YouTube video of looks like such a thing.

The way you describe your 'plan' smacks of you having all these grand concepts but no clue as to how they are actually implemented or the problems you'll face. You just say "I'll get a neural net to do that", as if that's a sufficient method. Given you can't do any mathematics I doubt you'll be doing anything other than just implementing neural net models others have devised so you won't be rewriting the paradigm. At the moment you sound like you're spending all your time on the pointless stuff, the arty crap anyone familiar with CAD could do. It's like someone spending ages finding a really good font to write their (so they claim) Nobel Prize winning paper in and never actually getting around to doing the work itself.

 

You don't use mathematics, you move bits, which is more like moving particles. The game on Youtube is a standard Neural Network, I wanted to test out the standard version before I worked on a new version. Of course your mind reading abilities get you nowhere.

 

So you don't deny you haven't ever done anything other than implement things other people have devised for you. Yet you make all these grand claims. You haven't got any idea how difficult any of these problems are, you just assume you'll be able to solve them. You're all talk with nothing to say.

As for my abilities, mind reading or otherwise, they have served me well in several of the areas you mention.

 

I don't know why you always alter people's replies into what is in your head...

That reply isn't stated anywhere in my post. You are getting my Neural Network confused with my Universe Generator. My Universe Generator is the program that isn't allowed to use maths. My Universe Generator is the program that is the original program. Think Tank it says at the beginning is a test of a Neural Network program.

Here is the quote from youtube...

It says hope to improve.. which means write another program.. my robot program.

 

Some ideas have been rehashed over and over and no matter how insane the idea is -- it seems all those who share it have a certain underpinning. This one in particular annoys me; "I will write an AI Neural-Network", or a "Learning machine". Often the ideas are grandiose and are comprised of a mix of delusion and misunderstanding. People tell me often they're going to work on this but it's unbelievably difficult and even saying you're going to attempt it exudes an arrogance which rubs me the wrong way.

I've done a little AI at times using ML and/or Lisp, there's a huge disconnect between the underlying logical theorems and the projected goals. At best I've create applications which can follow logical paths influenced by dynamic variables.

If you would have said you were going to start with something small and explain how atomic rules or functional transitions could lead to learned rules, someone might have give your post credence.

As it stand, you come across like an arrogant jerk who cares more about proving your intelligence than becoming educated.

 

Think Tank was my small step as I have already explained...

http://www.youtube.com/watch?v=zBtt-iVqvFE

What I like about Think Tank is that I gave it independent movement of its Left track, and independent movement of its right track. I never gave it movement of both tracks to go in a straight line. To watch it go in a straight line towards a mine it is adjusting both tracks to match speed, and I just find that interesting. Think Tank was never meant to be all that great however. I don't really see how I could step up any slower. I don't think I have mentioned that the Tank is blind either. It scores for being near a mine, but doesn't have any input of the direction of the mines.

 

The one working thing you've managed to demonstrate is, by your own admission, a standard thing.

Except it does use mathematics. I've explained this to you not once, not twice, not thrice, but at least four times (and that's ignoring other times I mention norms to you).

To even draw a sphere you're using mathematics. This is why things like machine vision are highly mathematical areas of research. To find something like a line or circle in a picture it's non-trivial. The way those are done in actual image recognition programs is through the Hough Transform. Specifically its implemented through the use of fast discrete Fourier transforms, the projection-slice theorem and thresholding. And that's just to find lines!

I'm sorry you not only understand that you're actually using mathematics beyond 1+(-1)=0 but you're actively ignoring multiple explanations but you can't magically avoid that by sticking your head in the sand.

If you are really doing viable stuff you shouldn't need to lie like that.

So that's another "In the future I will do...." statement of yours. Funny how you've got all these claims and plans and nothing to show for it.

People hear "AI neural net" and they imagine sci-fi like stuff when infact a neural net is a nice concept but rather mathematically mundane in its basic implementation. But even that will be enough to baffle Pincho, who doesn't even realise using numbers and distances involves basic mathematics.

Likewise for me. If you have one task, one very clear and precise task, then you can knock up something to do it pretty well. When you start getting into more elaborate things, beyond just basic classification or pattern matching, things get very unpleasant. Even if they didn't the scalings aren't nice, the size of neural nets to do some otherwise basic tasks can be completely impractical. Of course Pincho doesn't know this, he hasn't got that far in his grand plan.

Sure, there's tons of really interesting things someone can look at to do with cellular automata. The area is so vast it's hardly been explored and many problems are still open. Anyone with basic programming abilities can write their own CA engine or just get a Java one from online. Then they can explore the relationship between the rules and structures which appear in typical systems. Plenty of researchers do just that, trying to develop wider understanding of how small rules lead to big effects. But Pincho isn't interested in that, he's aiming for literally everything.

No, Pincho cares more about proving to people he has the intelligence he thinks he has. He's already proven his actual intelligence to all of us, the complete lack of substance or results despite all his grand claims settled that long ago. What it now comes down to is how attached to reality Pincho is.

 

I don't draw a sphere though do I so you are wrong. There is a sphere on the screen to represent a particle, but the screen is actually black because it is a zero particle, and they aren't really visible. So what you see on the screen isn't the actual program. But if I ran it as just a black screen it would be pointless writing the program. So you are the one lying.

 

Good luck!

 

You still don't get it. The definition of a sphere is in terms of a distance. Distances are mathematical in nature. To endow a construct with distances is to add mathematical structure.

For example, what is the distance between \((0,0)\) and \((1,1)\)? If you use the Euclidean distance then its \(\sqrt{2}\). If you use the Manhattan distance it is 2. If you use the infinity norm then its 1.

So to even have the construct of 'sphere' means you have picked a notion of distance, in a vector space. In a space with coordinate (x,y) what is (x,y) + (x',y') ? In a vector space it is (x+x',y+y') but there's systems where that isn't true. Just because you don't realise all the properties of the system you're assuming are in fact assumptions and mathematical in nature doesn't mean you aren't using mathematics.

Your 'universe generator' involves intersecting spheres. Suppose you wanted to consider spheres of radius 1. Two spheres intersect if their centres are within a distance 2 of one another. Suppose one is centred at (0,0) and another at (1,2). They will touch in the infinity norm but not in the Euclidean or Manhattan distances.

All of these are mathematical things you've assumed. You don't realise it because you're so profoundly lacking in mathematical knowledge you don't realise the stuff you learnt in school is not the only take on things. You think you're knowledgeable but actually it's because you're so lacking in relevant knowledge.

It's a typical mistake of people who haven't seen just how much mathematics there is to assume that the stuff they've done in school is somehow fundamental. It seriously is not.

Now this is at least the 5th time I've explained this to you. You really have no excuse to continue claiming you aren't using any mathematics unless you're deliberately trying to be dishonest.

 

I don't use sphere like that. I use them like this...

Infinity = Whatever your computer can cope with
Dim(Infinity,Infinity,Infinity,Infinity,Infinity,Infinity)

Then I just fill the Dims. I don't need the actual sphere shapes themselves. Sphere are made from smaller sphere, made from smaller sphere, made from smaller sphere.. infinitely. So I only need the locations. Then you have propagation. The only reason I draw the sphere to the screen is so that you can see them. They are actually infinite sphere constructs to obey Newton's Law that all actions have opposite reactions. If the universe is infinite, then the opposite reactions are infinite regression. To complete the infinite regression you also need negative infinite regression. You can get the locations just with the number 6, so that part is easy. Move 6, scale +6, scale -6... the next particle down is also 6, you don't need to change anything, just the scale of 6. Like a human uses feet, if you scale a human down his feet are smaller, so his scale is still in feet, but smaller feet. He is still 6ft tall no matter what size you shrink him to. In this way computers can replace impossible mathematics. But particles will always use no logic apart from staying apart from one another, and that logic will always be 6.

 

I cannot tell if you're deliberately trolling by pretending to be deluded and nonsensical or if you're actually that deluded and nonsensical. Either way your posts are becoming just gibberish, you're not even forming coherent syllogisms.