Artificial Intelligence

Post Reply New Topic

Humans can invent a pocket calculator. Can a pocket calculator invent a human? Some vastly superior that is.

ruveyn

Last edited by ruveyn on 22 Dec 2009, 9:10 am, edited 1 time in total.

the only sentence up there I can fully agree with. sorry.

I frankly disagree.

If we examine the human machine, many operations done on an "intuitive" or "emotional" level follow a binary formula. The problem is that these operations don't adhere to traditional "mathematical logic."

Logic applies in all circumstances....the difference is that in some circumstances, the rules are different.

1+1 = 2 in mathematical logic.

What happens in a universe where 1 actually = 3.

Then 1+1 = 6.

Logical, but a FALSE response if you are applying the wrong set of rules.

With computer-based intelligence, we try to distill human concepts into logical formulas that a computer can process by a mathematical logic point of view. Maybe we need to find ways to get the computer to solve formulas via different sets of rules?

ever heard of fuzzy logic?

I use it all the time.

No two humans would solve a problem the same way, where is intelligence?

We do keep things that work, but that came from one mind.

There are thousands of these threads of new concepts coming up at any time, which any one human may not have heard of.

Humans do make a series of yes no value related judgements and do not make others, to reach an action plan, not a solution. It works sometimes.

AI will only have the data base, a set of rules to follow, and no human does that.

Humans try, try, and try again, they are failure prone. AI would have a breakdown.

These are two incompatable systems.

This kinda reminds me of the game "Mass Effect" .

But, with respect to AI . . . well, as been seen, the whole psychological issue comes into play. Free Will, Intelligence, Conscience, Subconscience, etc are all things that have to be understood before applied.

Free Will: My take on this is that we have it (ie the right to choose), but its our options that are limited, which are determined by the situation. For example, as described by the movie "Road House," "If someone is pointing a gun at your head, you've only got two options: die, or kill the <expletive>." Course, it also depends on how good you are and even if you can come up with other options on short notice. Scenario determines options, you determine the one that suits you . . . and live with the consequences.

Intelligence: For me, this is the ability to acquire, maintain, and apply knowledge and experience. This leaves room for everyone (as far as a general understanding goes). If something is of high interest, a person will be overly intelligent with respect to it, but could/would be woefully ignorant of anything outside it (severity of interest pending).

Conscience/Subconscience: If you'll pardon the analogy, i view this as a kind of database. Conscience takes in and digs up information (experiences), and the subconscience is like a simulator that works off the database. How to explain this (since im into computers and theyre the closest thing to the brain at the moment) . . . taking the 1+1 example. We experience that 1 (a single object/entity) and another is brought in, we now have 2 objects/entities. So, whenever there is another situation with different objects, there will be 2. The former statement is obtained consciencely, the latter is application through the subconscious (i would say logic, but its merely the application of a pattern . . . granted, thats just my definition of it).

Effectively, the AI would need these components: to learn patterns, screw up (make mistakes), then reapply experiences . . . oh, and dont forget the Three Laws.

returning to when we were talking about ai as a technical field rather than (more or less) murky philosophical issues.... yes, that is something I've been dabbling in more recently---specifically a sort of application of information theory of continuous, active stochastic processes used to implement a form of artificial logic network. based on a recent publication of mine. very interesting stuff, that. using it for generative content and as a NaturalMotion Endorphine-type middleware (albeit open source) with the Blender Game Engine (which is not very good, as I am aware, but this is purely for demo purposes).

no problem there.
free will: an illusion as explained above. an artificial mind with a rich self representation and conscious cognition feedback like any human will feel to have a "free will".

Intelligence: a psychological pseudo-metric attempt for the fitness of a selected brain. In common language meaning a combined measure for knowledge, processing speed and other aspects of human thinking in real life application.

Conscience: the continuous integration of real world sensory input with a dynamic self representation leads to an effect felt as consciousness.

Subconscience: You cannot draw a hard line between conscious and sub-conscious processes. Actually we have a standing, high dimensional neuron spike wave, which dynamically morphs within the neural super-network called brain and multiple foci which move and morph within its substructures. Information subsided in regions which are not currently heavily activated are therefore not merged into the self representation part lying in the "working memory". This doesn't mean though, that these regions do not contribute to the overall process and may be activated again later on. So the brain "thinks about" (computes) a lot of things without us noticing. And when the results of some unattended process suddenly correlate enough with the rest of the activity, those patterns will become stronger again so that they can re-emerge into the self representation in working memory, resulting in the feeling that some "sub-conscious process" has done something.

Please note that these are extremely simplified explanations which can only be fully understood if one understands the actual way the brain works. We use these words to talk about the working of a biological brain, but these words were formed and attached definitions by people who had not the slightest idea about the brain at all. They do not really apply, so explaining them for me feels like an attempt to explain concepts from newtonian physics despite knowing about relativity and quantum mechanics to children, who don't even understand newtonian physics completely. (no offence intended)



I don't see it that way. You may feel to have many options but you will chose only one. And in reality this was the only option, as you obviously weren't able to really choose any other one.



Those definitions are not bad, but as I mentioned above are very irrelevant (due to the concepts themselves being irrelevant), because they ignore reality. Reality though says that the brain is a complex biological network structure with a high dimensional electric spiking software running on it.



No, AI needs a high dimensional structure from which patterns like a self representation can emerge like in the human brain. And it needs a connection to the real world like a body. Then behavior and personality will emerge.

And of course you know about the flaw in the three laws, otherwise read the robot books again

In Germany they built a six-legged robot based on a neural network system.
It learns different gaits, which it then uses to adapt to the terrain.

Here's the video and article:
New Scientist - Panic walking gets robot out of sticky situations

That kind of thinking would help. One of our Mars rovers is stuck in a hole. It was ground based control that sent it into the hole. A fairly simple education would tell it what places to avoid, even if the ground based human said do it.

A sense of self preservation, the willingness to refuse orders, could make a better machine, but a dangerous one. Robot Panic sounds like something I would avoid.

not very impressing. look at this one:

evil military robot - walking in real terrain and even on ice

thats impressing! its actually nearly frightening. still, all this has nothing to do with artificial intelligence in spite of such work being labelled so. They have succeeded in building artificial spinal ganglia, this is a far way from and very different to building a brain.

Yeah, what ever happened with NASA following asimovs 3 laws of robotics?

Does anyone know of any good grad schools or professors for Artificial Intelligence/Cognitive Psychology?

At Imperial College London iCub, a robot toddler, is getting a makeover to make it/him/her even more like a real toddler. iCub will get tiny hands, new legs and a new brain. iCub was built to test theories on how children think, learn and develop. They are also planning to give iCub tactile sense, so iCub will be able to grasp objects, and to give iCub skin.

Here's a video and a short article:
NewScientist - Robot toddler gets an upgrade

Has it been asking about Sarah Conner?