Anyone interested in a VB.net and/or C# project for creating real AI then mail me at da_rikard@hotmail.com and i shall tell you more about the CES Project and the SC-11 simulation.

BMW-Guy,

I'm a programmer of VB, Java, and many other languages. If you have any specific questions, I suggest you join up at www.asp.net (microsoft's .net forums), www.vbforums.com or www.webdesignforums.net (i moderate there).

PS - what's up, man? :D

Anyone interested in a VB.net and/or C# project for creating real AI then mail me at da_rikard@hotmail.com and i shall tell you more about the CES Project and the SC-11 simulation.'Real AI' in VBnet is just a funny group of words that should never be used together.

As for 'real AI' in C#... well it's more likely... but I have no idea what you mean by 'real'

Perhaps they mean "Real Artificial Intelligence" as aposed to "Something called Artificial Intelligence thats just a Bot(Autonomous piece of code)", however I don't think a real AI would be programmed in any conventional language, it would require at least assembly level and the entire manipulation of intelligence would have to be right down the very logic gates that the understanding of the language is constructed.

(Afterall it's said we store and process information at the genetic level, this is what it would mean for a computer to assimilate the same)

The problem is that we don't REALLY know what intelligence is. We know it has something to do with the brain. Beyond that we are fairly clueless.

"Afterall it's said we store and process information at the genetic level"
I don't think this referes to the knowledge we is used by our 'intelligence'

I have been programming since i was 7, i think i can determine which language i wish to use myself.
However, there is not language which will optimally give my AI concept justice so we will create the CES programming language later on when we get paid for our work.

What do i mean with Real? Well, human intelligence. I have a new concept with which i really can create a human replica, the beauty is that i do not need 10 super computers, i do not even need 1. A normal PC would do just fine for my AI, but it would need to be modified a little so that it suites our AI a lil better.

k, the SC-11 simulation.
We will start a artificial lifeform up in a room (will be built as a robot in the near future). We will start it up with no knowledge. (which will prove how adapable it is).
It will have to figure out how to use the key on the door. It will do this on the first try and do you know why? Because it is as smart as you and I.
When it figures that out it will get to room 2. In room 2 it will have to build something more advanced. I thought first of a electro-magnet so that it could shoot a hole in the wall. But that might harm the lifeform when it is a robot so i will have to change room 2.
In conclusion, it is as smart as a human. my algorithm is universal so it would solve almost any problem.
Don't believe me? Nah, i am used to that. However if you are interested then mail me at da_rikard@hotmail.com and ill tell you more.

We may be hijacking this thread... but I think Stryder can split it if he feels that is so.

We will start a artificial lifeform up in a room (will be built as a robot in the near future). We will start it up with no knowledge. (which will prove how adapable it is). It will have to figure out how to use the key on the door. It will do this on the first try and do you know why? Because it is as smart as you and I.Ok... you are skipping some steps. Basically you are saying:
1) I will build a robot
2) I will write a program
3) The program will have human intelligence

Could you please explain what is involved in step 2. I eat up anything having to do with AI research and have written a few programs involving it (mostly parsers, NLP, and non-recurssive path-finding).

I'd bve interested to see how you plan to implement human intelligence in a computer, when we don't know what human intelligence actually is.

To use the field's terminology, you're talking about 'strong AI' and 'weak AI.' Strong AI is an AI that can reason and think for itself, and is very likely sentient. Weak AI is what we have had to date, where the AI can only act as if it's intelligent by emulating human behavior.

I don't mean to be overly harsh, but I'll believe this magical artificial life form when I see it. For one thing, I see a large problem here:
We will start it up with no knowledge. (which will prove how adapable it is).
It will have to figure out how to use the key on the door. It will do this on the first try and do you know why? Because it is as smart as you and I.
On the first try? It must be pure chance if it has no knowledge. The only reason you or I would try to use the key in the lock is because we've been trained to use keys in this manner.
Or do you mean that you'll give it a key, point it at the door, and tell it to figure out how to use the key on that door? That sounds like knowledge to me.

I have been programming since i was 7, i think i can determine which language i wish to use myself.

Regardless of how long you've been programming, there are many things you either cannot do in VB, or take quite a bit more effort to accomplish than it would with another more powerful language. VB has its place, and that place is on the Windows desktop.

my algorithm is universal so it would solve almost any problem.

That's quite a strong statement. I get the sense that your concept is still in the idea stage and you haven't produced anything yet. I would be very intested to hear if anything develops once you get it to a working stage, as I think you'll probably run into some problems that you're not expecting. I'm not trying to discourage you, I'm just trying to point out that you're probably in for more hurdles than you think.

well actually we have been doing this for almost a year now so we are in the developing stage but you are right, i have improved the AI on the path.

We do not need super-computers to develop our AI, nor do we need the extra power that C++ could give us so id rather use the language that i find the simplest.

No, i will not give it a key and point at the door but it will have the ability to interact. Please note that i do not have to tell it anything about its world only that if i do not then it will be next to impossible to read the output.
It cannot see color, shape. It only sees text and even that is not neccesary. Can it be integrated in the real world? Yes, but it requires some of the most advanced robotic eyes for that.
All that it really needs to get as an input from the SC-11 simulation is to know when a state changes. Etc, when a door is unlocked instead of locked. Just things that will be able to detect.
I don't want to say much about the AI and I do not want people to believe me, actually i want to prove people wrong but the truth is that ill need someone more with lots of time who is a grand VB.net/C# programmer. The others do not have much time so ill need one more atleast. Mail me and i shall tell thee more.

Persol,
I'd bve interested to see how you plan to implement human intelligence in a computer, when we don't know what human intelligence actually is.
Not knowing what to implement, is not a fatal problem if you do know what it should mimmick, e.g. a human. The Turing test (http://en.wikipedia.org/wiki/Turing_test) circumvents our inability to accurately define intelligence with a comparison on a feature of intelligence (to fool around a questioner) between humans and software.

Baal Zebul,
Please note that i do not have to tell it anything about its world only that if i do not then it will be next to impossible to read the output.
Why? How is the readability of the output related to its initial knowledge of the world?

All that it really needs to get as an input from the SC-11 simulation is to know when a state changes.
So what state changes do you envision? How do you describe them? How is your artificial life form able to interact? Does it have the ability to actively change states? If so, should i see it as a MUD like command language (e.g. "use key on door" type of commands)?

I don't want to say much about the AI and I do not want people to believe me, actually i want to prove people wrong but the truth is that ill need someone more with lots of time who is a grand VB.net/C# programmer.
Well, if you want people to get enthousiastic about your project, you at least need to be able to draw a general outline of how your concept works. As Persol and Malkiri illustrated, there are some serious issues you have left unaddressed in your explanations. Without addressing them properly, I can imagine that it will be quite difficult to inspire talented programmers to spend time on a project lacking a concrete basis.

Baal Zebul, I would suggest a cheaper alternative to Robotics, Try using VR. (Virtual Reality) this way you can generate objects that you interact with at the same level as the intelligence your trying to create.

You could create and name an object a chair, and the Intelligence would then decide what it should remember as a chair, but do note with human intelligence a definition of chair would have to be repeated many times to ascertain what a chair actually is. In fact occasionally our primitive development when that young would cause other things to be refered to as chair.

(Like parents teaching a baby how to differenciate, where a mother is with the baby more often and the irony of the first word it knows being "Daddy".)

Baal, perhaps you should contact Jim Wightman, who I believe has beaten you to solving the AI problem with his nannybots.
Or perhaps you ARE Jim Wightman.

mouse,

Why? How is the readability of the output related to its initial knowledge of the world?

Stryder gets it but not you. He said the thing about calling the chair a chair. Well that is all i tell it in SC-11 (even though it does not even have to know that). But what is easiest to understand when reviewing the output, That it says "Key" or "2a42hi412sd" or some other randomly self-created "word"?

So what state changes do you envision? How do you describe them? How is your artificial life form able to interact? Does it have the ability to actively change states? If so, should i see it as a MUD like command language (e.g. "use key on door" type of commands)?

Yes, you should. Somewhat like that. Let me ask you, how do you know that you have opened a door? Maybe you have experience or may be you are lucky. Maybe you see that the lock has moved or maybe you hear the click. There are tons of status changes. However my AI only needs the status changes when it has no information and i have designed SC-11 accordingly were it survives with the absolute minimum of information.

Well, if you want people to get enthousiastic about your project, you at least need to be able to draw a general outline of how your concept works. As Persol and Malkiri illustrated, there are some serious issues you have left unaddressed in your explanations. Without addressing them properly, I can imagine that it will be quite difficult to inspire talented programmers to spend time on a project lacking a concrete basis.

You know, i have been to a lots of places. I in all of those places used a technic that i learned from when trying to rally artists for my game developing crew. What i do is that i tell everybody that i am better than them (in some way) and the result are (from the noobs) that they call me insane or a liar whilst the professional contact me, join me and work till they have proved that they are better than me.
You might want to remember this for future reference. But ill not do that here at SciForums because people here answer posts so quickly and most of the time have something intelligent to say that id not want to risk my "future" in this forum.


And Stryder when it comes to VR, well just say that i have an other invetion there which is lets say 5 years before its time. We plan on selling that too but the cost is somewhat large. We were first to build The Republic but that was obsolete when i thought a lil more about it. Then came The Republic 2 (TR2), a good idea since we would start them with no knowledge as in SC-11 but there were too few dead ends in TR2 so i dropped it rather than modifiying it. SC-11 was the next step, (we have changed it now thanks to a "friend" of mine) and it will be put in a maze with no knowledge (sort of like a RPG game) where it tries to get from point A to point B and along the maze there are obsticles such as the door issue.

You could create and name an object a chair, and the Intelligence would then decide what it should remember as a chair, but do note with human intelligence a definition of chair would have to be repeated many times to ascertain what a chair actually is. In fact occasionally our primitive development when that young would cause other things to be refered to as chair.

Yes, that is what i plan on doing and have done already but thanks for saying "but do note with human intelligence a definition of chair would have to be repeated many times to ascertain what a chair actually is. In fact occasionally our primitive development when that young would cause other things to be refered to as chair." I had not thought of having empirc data for objects also (yet), thank you indeed.

I would suggest using www.sourceforge.net as a place to launch a project, since it would give you enough space and a website to allow you to generate a Project Goal and statement that people can refer to.

Since you are claiming to be so many years (I would estimate over 50) in advance of the bleeding edge of technology, could you please give us some additional details.

What architecture for learning language are you implementing in your A.I. construct?
If it has no language, how does it reason?
How does it solve implicatures, i.e. I know X, I know Y, therefore I infer Z.
Does your construct have access to any external ontologies/dictionaries/corpus?
What are its sensory abilities, its inputs?
What problem solving algorithms does it implement?
How are you planning on translating its output?
How are you planning on representing objects?
How are you planning on resolving abiguity between different senses of a given concept? (example: he knocked on the door, versus he walked through the door)
What data structures does it use to hold knowledge?
Is it evolving?
Does it use rules, finite state machines, ATMs?
Give us a clue!

If you do not give more details, you cannot blame people for assuming that you are an inventor only of wild ideas. I work on a tiny corner of the A.I. problem, a subset of a subset. My belief (and a well-founded one) is that A.I. is a huge unsolved area of study; a problem to be chipped away at, not solved in one great leap.

Stryder gets it but not you. He said the thing about calling the chair a chair. Well that is all i tell it in SC-11 (even though it does not even have to know that). But what is easiest to understand when reviewing the output, That it says "Key" or "2a42hi412sd" or some other randomly self-created "word"?
Fine, what process would get your AI decide to produce a random string for something? Does it make an inventory of all that it "sees" in its virtual world, and attach labels to it if not done by you already? Which brings me to another question, does your system take into account a line-of-sight, does it register objects hidden behind a wall?

Yes, you should. Somewhat like that. Let me ask you, how do you know that you have opened a door? Maybe you see that the lock has moved or maybe you hear the click. There are tons of status changes.
We humans have figured out which status changes are important, and which we can filter out. That took time to learn, and moreover, we have a mechanism enabling us to learn in the first place. What mechanism do you envision for your system to know what status changes are relevant to complete its goals and which can be disregarded?

However my AI only needs the status changes when it has no information
What do you define with this information? Is it its goal list, its task list, its list of objects it can "see", a list of previous status changes, or a combination of these?

i have designed SC-11 accordingly were it survives with the absolute minimum of information.
How do you qualify a minimum of information? Do you mean that if you tell it that there is a room, a door, a key and that its sole purpose in life is to get through the door, that this is the minimum information needed for it to succeed? If so, how does it know that a room consists of walls, a floor and a roof? Does it have access to a dictionary defining a room as such? And what about the definition of a door, a wall, a floor and a roof? How does it know that it can not move through walls, floors, roofs, closed doors? Does it just try to move through one of these and remembers the failure of doing so? If so, how does it generalize from one wall to another? Does it assume that if a certain action fails on a object of a certain class, that it will fail on every other object of that class? If so, how do you account for exceptions and dependencies. If it discovers that a key can be used to open a door, it could incorrectly conclude that every key can open all doors. What mechanism is built in to counter or refine previously made assumptions?

And Stryder when it comes to VR, well just say that i have an other invetion there which is lets say 5 years before its time. We plan on selling that too but the cost is somewhat large. We were first to build The Republic but that was obsolete when i thought a lil more about it. Then came The Republic 2 (TR2), a good idea since we would start them with no knowledge as in SC-11 but there were too few dead ends in TR2 so i dropped it rather than modifiying it. SC-11 was the next step, (we have changed it now thanks to a "friend" of mine) and it will be put in a maze with no knowledge (sort of like a RPG game) where it tries to get from point A to point B and along the maze there are obsticles such as the door issue.

Dropping names and acronyms out of context doesn't make your project sound better. Please, at least explain what "The Republic" and "SC-11" are if you're going to refer to them.

Another problem with claiming your AI is as intelligent as a human is that you're interpreting the input and output on its behalf. This:

You're in a room.
There's a door here.
There's a key on the floor.

is not the same as sitting a robot with sensors of some sort in an empty room with a door and a key. I'll even assume it can identify the key and door, which is not a trivial problem. There's a large difference between your AI replying "use key in keyhole in door" and having the AI-in-robot pick up the key, put it in the keyhole, and turning it.

Lastly...if you're willing to divulge details to people if they email you, why are you reluctant to post more of them here?

Forgive me for being so breif about my AI, i am highly paranoid and since i do not want any competition i try not to explain much about it. I barely trust PGP to share my ideas.


Stryderunknown,

I would suggest using www.sourceforge.net as a place to launch a project, since it would give you enough space and a website to allow you to generate a Project Goal and statement that people can refer to.

Perhaps, but i already have lots of other places where i can upload my website. It is not complete yet but ill look into using sourceforge.

What architecture for learning language are you implementing in your A.I. construct?
If it has no language, how does it reason?
How does it solve implicatures, i.e. I know X, I know Y, therefore I infer Z.
Does your construct have access to any external ontologies/dictionaries/corpus?
What are its sensory abilities, its inputs?
What problem solving algorithms does it implement?
How are you planning on translating its output?
How are you planning on representing objects?
How are you planning on resolving abiguity between different senses of a given concept? (example: he knocked on the door, versus he walked through the door)
What data structures does it use to hold knowledge?
Is it evolving?
Does it use rules, finite state machines, ATMs?
Give us a clue!


Who said that it does not use language?
It uses the CES language, which is similar to english with some grammar modifications, i believe that i have already said that.

From NN i have created sort of a mixture between Trail and Error and Back-Propagating systems. (Back-Propagating might be a little clue, Just a hint)

In the real world it would have eyes that can see 3D, they should be able to make a 3D view of an object and apply a texture to it. (This would allow it recognize ojects even if it only has seen bits of it) It would also be able to zoom in and out. This is the minimum but in military application it would of course also have night and heat vision and it would probably be able to detect hidden weapons and see through walls too.
In the digital world it does not have to see more than text, however in order to create more advanced simulations one might give the abbility to see color and shape.
I do not have to translate the output. I do not have to translate the input.

This is the best i can say, it is pattern recognition on multiple levels but structured as neural nets, using a mixture between Trial and Error and Back-Propagating systems, this gives a what you would get out of Genetic Algorithms and Expert Systems. It creates its own natural language based on interaction and knowledge. In short, it is human or next to human.
(I can only put it in the previous AI terminology cause then people cannot understand how it works fully.)

mouse,

Fine, what process would get your AI decide to produce a random string for something? Does it make an inventory of all that it "sees" in its virtual world, and attach labels to it if not done by you already? Which brings me to another question, does your system take into account a line-of-sight, does it register objects hidden behind a wall?

Yes, almost like that.
Inh military application it might be able to see through walls but that is not standard. However it will always try to examine the objects more and more.

We humans have figured out which status changes are important, and which we can filter out. That took time to learn, and moreover, we have a mechanism enabling us to learn in the first place. What mechanism do you envision for your system to know what status changes are relevant to complete its goals and which can be disregarded?

Well, what is important? If you only will open doors then you dont have to know much. However if you are to repair door then you might need to know more, right?

What do you define with this information? Is it its goal list, its task list, its list of objects it can "see", a list of previous status changes, or a combination of these?

A combination and a little more to it.

How do you qualify a minimum of information? Do you mean that if you tell it that there is a room, a door, a key and that its sole purpose in life is to get through the door, that this is the minimum information needed for it to succeed? If so, how does it know that a room consists of walls, a floor and a roof? Does it have access to a dictionary defining a room as such? And what about the definition of a door, a wall, a floor and a roof? How does it know that it can not move through walls, floors, roofs, closed doors? Does it just try to move through one of these and remembers the failure of doing so? If so, how does it generalize from one wall to another? Does it assume that if a certain action fails on a object of a certain class, that it will fail on every other object of that class? If so, how do you account for exceptions and dependencies. If it discovers that a key can be used to open a door, it could incorrectly conclude that every key can open all doors. What mechanism is built in to counter or refine previously made assumptions?

Well, it will not know that there is a roof since the roof in no way could affect the simulation. It will have no dictonary, it will create a dictionary.
Yes, you are right. It can think that a key can open a door when the truth is that that key just openes one door. That is proved with empirical data, so if it solves the problem once and fails two times then it has learned that that key just worked for that particullar door. It could also think that the empirical data proves that the key can not open any more doors but that will not happened since it has worked once, so it will know that that key will work on that door.

malkiri,

Dropping names and acronyms out of context doesn't make your project sound better. Please, at least explain what "The Republic" and "SC-11" are if you're going to refer to them.

Sure, this is a copy from our website. I have not fully completed the SC-11 part yet but i guess that does not mean anything to you any way.


"The Republic (TR)

The Republic is a simulation of a medieval society with 19 inhabitants, all with their specific capabilities and their tasks. The idea with the republic was to show the adaptability of the CES AI in order to prove not just its human like intelligence but also show that it was as grand at adapting to harsh environments as bacteria.

What really failed the Republic was that it was created in order to make sure that if one inhabitant failed then the whole society would fail within a short time if it could not adapt. However, The Republic was too pre-programmed in the sense that it was created within boundaries that always would give a positive result.



The Republic 2 (TR2)

In order to achieve more human results with a higher error percentage it was decided to create the Republic 2, a world with 3 individuals, Adam, Eve and Jesus.

Adam was the father, the strong man suited for physical activities such as cutting wood and hunting whilst Eve had the capabilities of foraging and cooking. Since the idea with the Republic 2 was that they should be started up with no knowledge what the CES team wanted out of TR2 was for Adam and Eve to solve all problems in the world and then teach baby Jesus all the knowledge they had so that the society could go on. The disadvantage once again was that the Republic 2 was too limited, too little dead ends that the AI would try to solve before deciding to move on to the next problem. A slight modification in the world we built for the ALF's could correct that mistake but the glitch was too excessive since the amount of data it required was overlooked. The amount of data that TR2 would require was comprehensible to some of the most advanced AGI creations and by doing so the CES Project had failed in their beliefs.



SC-11

When the TR2 simulation was obsolete SC-11 was the next concept that would be undertaken rather than continuing the trend of creating a better and more advanced republic simulation.

There are no disadvantages to the SC-11 simulation except perhaps that it cannot simulate every problem that might occur in the real world. "


is not the same as sitting a robot with sensors of some sort in an empty room with a door and a key. I'll even assume it can identify the key and door, which is not a trivial problem. There's a large difference between your AI replying "use key in keyhole in door" and having the AI-in-robot pick up the key, put it in the keyhole, and turning it.

Why not? The robot will reason in the same manner as the simulated robot. Turning is an action and instead of sending the output to a human viewed text it just sends it to the parts in its body that are affected in the Turn command.
Key and Door are merely object (no matter what they are called), they just give X, Y and Z cordinates for the robot to use when calibrating its Turn command.

This will probably sound even worse but i do not care, being on the safe side is always better.

In the real world it would have eyes that can see 3D, they should be able to make a 3D view of an object and apply a texture to it. (This would allow it recognize ojects even if it only has seen bits of it) It would also be able to zoom in and out. This is the minimum but in military application it would of course also have night and heat vision and it would probably be able to detect hidden weapons and see through walls too.


Your system can consistently identify arbitrary objects from an arbitrary viewpoint, particularly from a view of only a portion of the objects?

I do not have to translate the output. I do not have to translate the input.

As I mentioned in my last post, you are translating both the input and output when you run the AI in a text simulation.

Well, what is important? If you only will open doors then you dont have to know much. However if you are to repair door then you might need to know more, right?


Are you saying that your AI doesn't need to discard apparently irrelevant information as it might be useful later? This misses the point of what mouse was saying. I hear keys tapping from other cubicles while I sit at my desk all day long. I don't remove this fact from my knowledge base. However, I also don't react to it - I simply ignore it. I think the question was more along these lines - how will your system know which environmental changes to ignore?

Why not? The robot will reason in the same manner as the simulated robot. Turning is an action and instead of sending the output to a human viewed text it just sends it to the parts in its body that are affected in the Turn command.
Key and Door are merely object (no matter what they are called), they just give X, Y and Z cordinates for the robot to use when calibrating its Turn command.

Because it needs to know which parts of its body to move and how to move them. Since the robot is given no knowledge ahead of time, this means it not only has to figure out that putting the key in the door, turning it, and opening it will bring it closer to its goal, but it also has to figure out how to reach down, pick up the key, move to the door, put the key in the keyhole, turn the key, grasp the doorknob, and open the door. All this information is not required when the AI will simply output "get key," "open door with key."

I did not say it didn't use language. I only said, if it does, what mechanisms does it have for learning it? Do you use the 'principles and parameters' approach to language learning, are you a connectionist, a lexical functionalist? What type of grammar are you using?

Baal Zebul
[...] i do not want any competition i try not to explain much about it.
You are planning a financial gain from your project? If that is the case, what prospects do you offer your potential programmers?

It uses the CES language, which is similar to english with some grammar modifications, i believe that i have already said that.
If it is similar to English, how do you propose dealing with the ambiguities that leda noted a few posts earlier?

From NN i have created sort of a mixture between Trail and Error and Back-Propagating systems. (Back-Propagating might be a little clue, Just a hint)
This is a bit confusing. The back propagation algorithm is based on trial and error: it is a method of adjusting weights by comparing the output of the neural network with a sample set. How do you propose a mixture of those two terms, if they are already linked with each other?

In the real world it would have eyes that can see 3D
See 3D? How does that work? Even our own eyes only see in 2D, the light falls on a flat retina obviously registering a 2D version of the world. If i remember correctly, the perception of depth is introduced in our brain, where, among other factors, the different visions from left and right eye are used to create the illusion of depth. Maybe someone in the biology section can give you a more accurate or detailed description of this process.

This is the best i can say, it is pattern recognition on multiple levels but structured as neural nets, using a mixture between Trial and Error and Back-Propagating systems, this gives a what you would get out of Genetic Algorithms and Expert Systems.
Multple levels of what? Layers in the neural network? First of all, I'm not sure if terminology is correctly applied here. As I've explained above, I can not envision a mixture between trail and error and back-propagation systems. Secondly, if such was possible i can not see how the result of it would be equivalent to genetic algorithms and expert systems. Both of these are approaches to completely different areas. Genetic algorithms is good for optimization, expert systems traditionally work on a more concretely defined rule set. What you are claiming to implement, human intelligence, is neither a pure optimilisation process nor a feasible with a concretely defined rule set.

It creates its own natural language based on interaction and knowledge. In short, it is human or next to human.
Humans learn a language by mimicking those who already mastered language. Who or what is your system mimicking? You?

I can only put it in the previous AI terminology cause then people cannot understand how it works fully.
I'm not sure if you are using AI terminology in the manner as I've understood it.

Well, what is important? If you only will open doors then you dont have to know much. However if you are to repair door then you might need to know more, right?
Yes, but how does your system know or discover which pieces of information are relevant and which are not? How does it make a correlation between an action and a state change? If they happen after eachother within a certain time frame? If so, how do you determine what is the most suitable time frame?

Well, it will not know that there is a roof since the roof in no way could affect the simulation.
But if the roof is there, and the bot can "see" the roof, why would the bot ignore it? What rule in it tells it that it should use keys, but that it should not try to use roofs? How ever silly the last piece of the question sounds, a bot without no initial knowledge what so ever, has no idea if a roof is relevant with regard to fullfilling its mission.

It will have no dictonary, it will create a dictionary.
Ok, how will it describe a wall? Attaching simply a random label is not creating a dictionary. If it is to be of any use, it should recognize what makes a wall different from other objects, like e.g. the floor.

Yes, you are right. It can think that a key can open a door when the truth is that that key just openes one door. That is proved with empirical data, so if it solves the problem once and fails two times then it has learned that that key just worked for that particullar door.
Why? There are other conclusions concievable. E.g. why not conclude that the two doors which did not open to the key appear to be not functioning? Given its lack of knowledge about the reliability of doors and its situation (a room, three doors and a key) it can not choose which hypothesis is the correct one or even assume which is more likely. How do you propose to handle this?

The robot will reason in the same manner as the simulated robot. Turning is an action and instead of sending the output to a human viewed text it just sends it to the parts in its body that are affected in the Turn command.
Key and Door are merely object (no matter what they are called), they just give X, Y and Z cordinates for the robot to use when calibrating its Turn command.
No, the real world is significantly more complex. Simple object recognition is not a trivial matter. How to recognize the key from the texture of the floor? How to recognize the key from different angles? And I'm not even speculating about the complexity in getting the key inserted in the correct manner in the door's key hole.

This will probably sound even worse but i do not care, being on the safe side is always better.
What safe side? You are worried that people are going to patent your ideas? I for one am not making a run on the patent office just yet. At this point, I can only see problems with your approach, rather than innovative solutions.

Baal,

Seems you first projects are usually refered to as "Ant Farms", namely worlds where multiple intereactive elements strive to survive and is based on how some people use to (and probably still) make ant farms in the real world.

I've seen a few older concepts when it was first starting to be used, like Bacteria's growth in a petri dish which was a kind of Chaos demo in itself, since it would seed differently based upon which square you clicked in a 20x20 grid, you could even seed more than one square. It followed the rule that from your seeds it would grow to the squares that neighboured it's sides, once that growth occured, the squares surrounding it would starve the bacteria where the seed occured, and then that dead bacteria area would become food to be eaten by the bacteria again. When it got to the edge of the petri dish it obviously reached an equillibrium of growth an decline. (which exists when you have closed systems)

I personally actually do want to get around to scripting a decent "Ant farm" for my own project, which involves creating a Virtual world where characters can be interacted with at the same level they compute at. (Namely you can see an object, so can the character. The object is defined a name, which both you and the character understands is the object. You can examine the object to which the character can examine the object. ... Move the object etc.)

However if such life is breathed into the characters in this way, will people still be able to play such games as Grand Theft, in the knowledge that the little old lady they ran over actually lived???

This doesn't approve very well thought out. You have mentioned that you are going to do a dozen or so things which require huge problems solved that people working on AI for years have not been able to solve....

...and you plan to do this with a group of random programmers who most likely have very little 'real' AI experience...

I personally actually do want to get around to scripting a decent "Ant farm" for my own project, which involves creating a Virtual world where characters can be interacted with at the same level they compute at. (Namely you can see an object, so can the character. The object is defined a name, which Move the object etc.)Now this sounds more realistic and interesting. Many games and the like seem to already follow the "both you and the character understands is the object. You can examine the object to which the character can examine the object".

Something else I should mention is that Baal's suggestion might not be as absurd as you might think, for instance take a look at http://www.visual-prolog.com/

A discussion on prolog was raise in the forums previously but my understanding is that prolog wasn't just used as a language with syntax created for problem solving and solution making, it was about creating something that had the ability to make decisions.

It lacks the functionality of C++ but is still utilised in this area, so it suggests VB can be used in the same way too.

It is suggested that prolog can be used to do that... but it hasn't actually been done. All it does is simplify dealing with the syntax... which still just leaves you with a bot.

Your system can consistently identify arbitrary objects from an arbitrary viewpoint, particularly from a view of only a portion of the objects?

that is not the everyday english that i know but i think it means that it can still understand the object even if it only sees a small part of the object, correct?

Well, it can always identify it even if does not have the whole picture however, it might not have enough info in order to do anything intelligent with it.

As I mentioned in my last post, you are translating both the input and output when you run the AI in a text simulation.

On the computer, no not really. However it will be pre-programmed so that it will get the right label on the items.
In the real world, i would translate the input and on the output it would be translated to robotic commands (a 1 here and a 0 over there perhaps)

Are you saying that your AI doesn't need to discard apparently irrelevant information as it might be useful later? This misses the point of what mouse was saying. I hear keys tapping from other cubicles while I sit at my desk all day long. I don't remove this fact from my knowledge base. However, I also don't react to it - I simply ignore it. I think the question was more along these lines - how will your system know which environmental changes to ignore?

Yeah, that i had not thought about actually. However it is easiliy changed. Just another topic for empiricial data. Let me ask you. If you were put in command of a nuclear plant without any knowledge, then how would you know which buttons to press if something happened on the first day? You would read a manual, ask somebody or guess, right? Why would my AI do anything else?

Because it needs to know which parts of its body to move and how to move them. Since the robot is given no knowledge ahead of time, this means it not only has to figure out that putting the key in the door, turning it, and opening it will bring it closer to its goal, but it also has to figure out how to reach down, pick up the key, move to the door, put the key in the keyhole, turn the key, grasp the doorknob, and open the door. All this information is not required when the AI will simply output "get key," "open door with key."

It is of course pre-programmed to some extent. Of course it needs to know how to move. I said that it was Real Artificial Intelligence and not Divine Artificial Intelligence.

I did not say it didn't use language. I only said, if it does, what mechanisms does it have for learning it? Do you use the 'principles and parameters' approach to language learning, are you a connectionist, a lexical functionalist? What type of grammar are you using?

Im sorry, i must have missunderstood you then.
Once again english beyond me, i would normally ask someone but i can't find anyone online better than me in english. Ill get back to this when i know the question.

You are planning a financial gain from your project? If that is the case, what prospects do you offer your potential programmers?

Well, we once had it up for discussion if we should create a AI programmer that could build us lots of OS's and sell them. I can only offer 0$ right now, however i am a fair man and we are 5 right now and we would each be entitled to 20% each of what ever, the more that join, the lesser percents each.
I have students in my team, (i am a stundet myself), i a professional AI developer on my team making 100,000$ a year. I am talking to other professional ai developers about them joing us whilst i have about 20 others wanting to join my team but they are all classified as people whom i cannot trust or people who know to little.

If it is similar to English, how do you propose dealing with the ambiguities that leda noted a few posts earlier?

Since i did not understand what leda said fully ill get back to this later too.

This is a bit confusing. The back propagation algorithm is based on trial and error: it is a method of adjusting weights by comparing the output of the neural network with a sample set. How do you propose a mixture of those two terms, if they are already linked with each other?

That i cannot say. It is not full back-propagating. Maybe i should had said that it uses a neural nets concept instead. This only confusses it i see.

See 3D? How does that work? Even our own eyes only see in 2D, the light falls on a flat retina obviously registering a 2D version of the world. If i remember correctly, the perception of depth is introduced in our brain, where, among other factors, the different visions from left and right eye are used to create the illusion of depth. Maybe someone in the biology section can give you a more accurate or detailed description of this process.

i know, but what i mean with 3D is that if you stand here then you do not see the same as you can see when you have moved one meter to the right. I have been making 3D games and when i think about 2D i think of a flat surface (a platform game) where someone walks on that surface, whilst 3D means that you can move in X, Y and Z. We only see a flat surface of course but i still think that it is a 3D environment.
A morphing sequence does not exist, it is just our brains that make it happen. So when you see someone transform from a human to an alien in movies then you have made that happen, but it is still called Morphing.

Multple levels of what? Layers in the neural network? First of all, I'm not sure if terminology is correctly applied here. As I've explained above, I can not envision a mixture between trail and error and back-propagation systems. Secondly, if such was possible i can not see how the result of it would be equivalent to genetic algorithms and expert systems. Both of these are approaches to completely different areas. Genetic algorithms is good for optimization, expert systems traditionally work on a more concretely defined rule set. What you are claiming to implement, human intelligence, is neither a pure optimilisation process nor a feasible with a concretely defined rule set.

The professional AI developer in my team has already called it "Neural Genetic Natural Language" He, just as me think that it is the core of AI, "mixing" all the other AI and combined creating real AI. (Not all concepts, ill not say which)

Humans learn a language by mimicking those who already mastered language. Who or what is your system mimicking? You?

No, reality. It creates the mimick, i never said that it would use the same words as we do. However if it was to have audio input then it could mimick the speakers of a conversation and use that data in many fields.

I'm not sure if you are using AI terminology in the manner as I've understood it.

I bet you that i would if i was american.

Yes, but how does your system know or discover which pieces of information are relevant and which are not? How does it make a correlation between an action and a state change? If they happen after eachother within a certain time frame? If so, how do you determine what is the most suitable time frame?

Yes, the only disadvantage is time. I press a button in room one, how the hell could i possibly know that it opened the door in room 5? I dont think there is any way actually except the concept of Origin Tracing, one of the earliy features of my design. However, why would anyone need to know that the button in room 1 opened the door in room 5? And if it really has too then it could use Origin Tracing but that would mean an substansical amount of trial and error really.

But if the roof is there, and the bot can "see" the roof, why would the bot ignore it? What rule in it tells it that it should use keys, but that it should not try to use roofs? How ever silly the last piece of the question sounds, a bot without no initial knowledge what so ever, has no idea if a roof is relevant with regard to fullfilling its mission.

Yeah, you are right. What i meant was that everything appear in text and the roof does not appear so it can't be used. However if it is just an empty room then it would try to do something with the roof but only once till it realises that it cannot do anything with it.

Ok, how will it describe a wall? Attaching simply a random label is not creating a dictionary. If it is to be of any use, it should recognize what makes a wall different from other objects, like e.g. the floor.

k, what is specific of walls? If you can only see text? Not colors, not material.

Why? There are other conclusions concievable. E.g. why not conclude that the two doors which did not open to the key appear to be not functioning? Given its lack of knowledge about the reliability of doors and its situation (a room, three doors and a key) it can not choose which hypothesis is the correct one or even assume which is more likely. How do you propose to handle this?

In SC-11 it would not. It lacks knowledge. In a human relica it would be dynamic depeding on previous patterns.

No, the real world is significantly more complex. Simple object recognition is not a trivial matter. How to recognize the key from the texture of the floor? How to recognize the key from different angles? And I'm not even speculating about the complexity in getting the key inserted in the correct manner in the door's key hole.

Well, it would make a 3D model of it and attach a texture to it. If you are talking about recognizing the object and not using it.

What safe side? You are worried that people are going to patent your ideas? I for one am not making a run on the patent office just yet. At this point, I can only see problems with your approach, rather than innovative solutions.

You cannot even see my approach and that is the way i like it. But you also asked me about what i would offer programmers that would join me, if it is money you want then i would have you sign a NDA first or most likely you would not even be in our team. I want to trust all my members and i do not have any NDA with them in order to create a casual environment.

However if such life is breathed into the characters in this way, will people still be able to play such games as Grand Theft, in the knowledge that the little old lady they ran over actually lived???

Have not thought about it in that way. Actually i think after what you just said ill not have the AI integrated in any games where they can die. Actually when i wrote the framwork for The Republic i instructed the others that they should abort the simulation if they would ever see that it was going to fail.
Stryder, we should chat sometime, you seem to be a trustworthy character.

...and you plan to do this with a group of random programmers who most likely have very little 'real' AI experience...

Random programmers? Well, all of us have experience in game development. We all know VB and/or C# (at least). We all agree that military application is the way that we should go even if Johan thinks that we should start with industrial application since it is a larger market.

A discussion on prolog was raise in the forums previously but my understanding is that prolog wasn't just used as a language with syntax created for problem solving and solution making, it was about creating something that had the ability to make decisions.

Almost any programming could be used since it is mainly just list processing however it requires the programming of the CES algo so i would not do it with Prolog even though it might be possible. Stryder, ill contact thee. Maybe we could have a chat this weekend, have you got MSN, ICQ or any chat application or should i just mail you?

It is suggested that prolog can be used to do that... but it hasn't actually been done. All it does is simplify dealing with the syntax... which still just leaves you with a bot.

Persol, you are the kind of person id not want in my team, i had to learn that the hard way. You are too negative for me, i did not have the complete algo when i started (it might have worked but it was not universal). I needed Vincent in my team who is a professional AI programmer. I asked him "what is missing" and he answered. I thought a lil about it and i solved the problems. That is what i do, i am a problem solver. The AI is probably not complete yet either but now i know that it is universal to atleast 85, 90 percents.

If you do not understand my question, how are you capable of producing an entity with language understanding capabilities? You MUST have some knowledge of linguistics and semantics in order to do this. In fact, you must have the most advanced knowledge of linguistics and semantic of anyone on the planet. This is my problem. You are basically talking about the equivalent of saying that you've built a time machine.

that is not the everyday english that i know but i think it means that it can still understand the object even if it only sees a small part of the object, correct?

Well, it can always identify it even if does not have the whole picture however, it might not have enough info in order to do anything intelligent with it.

I was making the statement as general as possible, to include every conceivable situtation. It can identify an apple that's 95% occluded? It can tell the difference between that apple and a picture of an apple that's 95% occluded? It can identify an apple viewed from the top, from the side, from the bottom?

If it can identify it, why wouldn't it have enough info about it to do something with it? By identify, I mean "determine the object's identity," not simply recognize that there is something there.


On the computer, no not really. However it will be pre-programmed so that it will get the right label on the items.
In the real world, i would translate the input and on the output it would be translated to robotic commands (a 1 here and a 0 over there perhaps)

On the computer - yes, really. You are translating the input, period. The AI doesn't have to deal with raw video or sensor data. You're telling it, "Hey, there's a key here."

Yeah, that i had not thought about actually. However it is easiliy changed. Just another topic for empiricial data. Let me ask you. If you were put in command of a nuclear plant without any knowledge, then how would you know which buttons to press if something happened on the first day? You would read a manual, ask somebody or guess, right? Why would my AI do anything else?

Will there be a manual for your AI to read? Someone for it to speak to? That leaves guess, or in other words, make a decision at random, possibly influenced by previous knowlege. Since it must consult its previous knowledge, it has to be stored in some fasion, right? And if the AI never discards any information, won't this knowledge base get pretty big? Didn't you say this AI will be able to run on a single home computer? I'm not talking about storage space, either...I'm talking about computing power, specifically the amount it'll take to search that ever-growing knowledge base.

Incindentally, I recall you saying, "It will do this on the first try and do you know why?" This doesn't sound like guessing to me.

It is of course pre-programmed to some extent. Of course it needs to know how to move. I said that it was Real Artificial Intelligence and not Divine Artificial Intelligence.

First of all, that's not what you said before. You said, "We will start it up with no knowledge." If that's not true, then you shouldn't claim it is.

Second of all - that's fine, let it know how to move. Are you going to also program it with the knowledge of how to insert and turn a key in a lock? If not, then it's not the same as a text simulation (recall that this line of discussion was regarding the difference between a text simulation and a real world test.) And if so, then in either case (text or real world), your AI will need to construct a sequence of basic movements that result in the key being inserted and turned. This is definitely not a trivial task.

On top of all of this - particularly if the AI is given no knowledge, why should it even expect that putting the key in the lock and turning it will open the door? Presumably you might say that this is the only option - there's a door, a key, and a keyhole. If I add more complexity to the problem, like adding various objects to the room, not necessarily key-like, why should the AI pick the key instead of the apple? Or the pencil?

I'm not saying that your AI will never be able to solve the problem of exiting the room. I'm saying you're making grandiose claims that assume no prior knowledge but produce divine results.


Yeah, you are right. What i meant was that everything appear in text and the roof does not appear so it can't be used.

k, what is specific of walls? If you can only see text? Not colors, not material.

Back to my previous statement - you're translating the input. You filtered out some of the information so that your AI doesn't have to deal with it.


However if it is just an empty room then it would try to do something with the roof but only once till it realises that it cannot do anything with it.

And again, you said the AI will get it correct on the first try.

No, the real world is significantly more complex. Simple object recognition is not a trivial matter. How to recognize the key from the texture of the floor? How to recognize the key from different angles? And I'm not even speculating about the complexity in getting the key inserted in the correct manner in the door's key hole.

Well, it would make a 3D model of it and attach a texture to it. If you are talking about recognizing the object and not using it.

To make a 3d model of an object, you first have to recognize it. "Recognize" in this context is not "determine the identity," but rather "distinguish from the background."

If you do not understand my question, how are you capable of producing an entity with language understanding capabilities? You MUST have some knowledge of linguistics and semantics in order to do this. In fact, you must have the most advanced knowledge of linguistics and semantic of anyone on the planet. This is my problem. You are basically talking about the equivalent of saying that you've built a time machine.

I have also been thinking about the time machine, but i have had no progress there yet :p
No, i do not know english as good as you do. I am not saying that i am going to preprogram the linguistics of my AI either, i am saying that it will create its own language. What i mean is that all words that i can affect it in will be in english. but the label that itt has learned all by itself in our world will be a string based on previous numbers of identified items. X = X +1 basically with a string before that number.

was making the statement as general as possible, to include every conceivable situtation. It can identify an apple that's 95% occluded? It can tell the difference between that apple and a picture of an apple that's 95% occluded? It can identify an apple viewed from the top, from the side, from the bottom?

I was making the statement as general as possible, to include every conceivable situtation. It can identify an apple that's 95% occluded? It can tell the difference between that apple and a picture of an apple that's 95% occluded? It can identify an apple viewed from the top, from the side, from the bottom?

thats the plan

If it can identify it, why wouldn't it have enough info about it to do something with it? By identify, I mean "determine the object's identity," not simply recognize that there is something there.

k, lets say that you identidy a nuclear missile. Well of course you can think of the usage area to launch it but you will most likely not think about maybe using the same concepts in nuclear plants now would you?
If it has identified the apple to 100% then it will know what to do with it. If it has identified it to 5 percents then it will probably do nothing with it.

On the computer - yes, really. You are translating the input, period. The AI doesn't have to deal with raw video or sensor data. You're telling it, "Hey, there's a key here."

Yes, somewhat like that. But i am not telling it that "This is the Key"
Im saying "Put this label on the identified item", Why is this a problem for you? I thought that i had explained it that it is easier to read "Key" than "9sa8dy8y42" or something (anything)

Will there be a manual for your AI to read? Someone for it to speak to? That leaves guess, or in other words, make a decision at random, possibly influenced by previous knowlege. Since it must consult its previous knowledge, it has to be stored in some fasion, right? And if the AI never discards any information, won't this knowledge base get pretty big? Didn't you say this AI will be able to run on a single home computer? I'm not talking about storage space, either...I'm talking about computing power, specifically the amount it'll take to search that ever-growing knowledge base.

My database system is very effective. In order to be able to open the door with the key it needs 4 entries. that is one kb.
A human replica would not be able to run on a normal PC (at the moment atleast)

Incindentally, I recall you saying, "It will do this on the first try and do you know why?" This doesn't sound like guessing to me.

Give me those 4 entries (1 kb) and ill have it solve that problem without guessing.
And i know the next question, what will those 4 entries be? That is the limit, i can talk about most things up to that point.

First of all, that's not what you said before. You said, "We will start it up with no knowledge." If that's not true, then you shouldn't claim it is.

I thought that was understood.

Second of all - that's fine, let it know how to move. Are you going to also program it with the knowledge of how to insert and turn a key in a lock? If not, then it's not the same as a text simulation (recall that this line of discussion was regarding the difference between a text simulation and a real world test.) And if so, then in either case (text or real world), your AI will need to construct a sequence of basic movements that result in the key being inserted and turned. This is definitely not a trivial task.

No, it will have a command but it will fill in the "blanks", it will update its X, Y, Z data into that command (this is for the real world simulation)
It will have to construct a sequence, yes. That is one of the key features of my AI.

On top of all of this - particularly if the AI is given no knowledge, why should it even expect that putting the key in the lock and turning it will open the door? Presumably you might say that this is the only option - there's a door, a key, and a keyhole. If I add more complexity to the problem, like adding various objects to the room, not necessarily key-like, why should the AI pick the key instead of the apple? Or the pencil?

Yes, it was decided after i first posted here that it will have the goal to get from point A to point B and therefore it has to open the door. We decided this cause otherwise it might try to do something with the apple and the pear first before it uses the key on the door. If it knows that the door is its objective then it will use the key. That still remains

'm not saying that your AI will never be able to solve the problem of exiting the room. I'm saying you're making grandiose claims that assume no prior knowledge but produce divine results.

Stunning results is the best way to become recognized id say.

And again, you said the AI will get it correct on the first try.

Yes, and it will, believe me it will. It will not get the maze right on the first try. Now that would be godlike. But the door it will solve.

To make a 3d model of an object, you first have to recognize it. "Recognize" in this context is not "determine the identity," but rather "distinguish from the background."

yes, identify is detect the object and create a entry about it.
That would allow it to use it (shape like) but not fully. Id like to say some more here but i cannot do that, i do not like competition especially if someone is smarter than me.

thats the plan

Quite impressive - can you faithfully identify an apple in this setting? How do you know it's not a mutant banana with an edge that happens to look like an apple?

k, lets say that you identidy a nuclear missile. Well of course you can think of the usage area to launch it but you will most likely not think about maybe using the same concepts in nuclear plants now would you?
If it has identified the apple to 100% then it will know what to do with it. If it has identified it to 5 percents then it will probably do nothing with it.

I misunderstood. I thought you were referring to a lack of information about the object that prevented using it, as opposed to a lack of information about the environment.


Yes, somewhat like that. But i am not telling it that "This is the Key"
Im saying "Put this label on the identified item", Why is this a problem for you? I thought that i had explained it that it is easier to read "Key" than "9sa8dy8y42" or something (anything)

I understand labeling and the reasons for it. That's not my problem. You can call it a "key" and the AI won't know anything new about it from that label. My problem is that you're recognizing (not identifying) the key for the AI. You're telling it that there is some object sitting on the ground in front of it. This information is not contained in the real world test.

Let me try a different angle. The text simulation does not contain visual data that the AI will need to process. It contains some descriptive strings. In the text simulation, the AI does not construct those descriptive strings - you do. In a real world test, the AI will construct those descriptions (in a different format, most likely). Therefore, you are translating the input.
The text simulation presumably does not describe everything in the room in the same detail as images of the room would. You'll describe the key in whatever detail you like. You'll describe the door. Will you describe the floor? The walls? Any other objects? Will you do it with enough detail to distinguish very similar objects from each other, no matter how small the difference? You'll probably end up working on the first room of the simulation for more time than you'll work on the AI itself. Because you're not giving the AI all the possible information, you're filtering the input.
Because you translate and filter the input in the text simulation, there are marked differences between it and the real world test. Therefore, your results from the text simulation will not indicate feasiblity in a real world test.

My database system is very effective. In order to be able to open the door with the key it needs 4 entries. that is one kb.
A human replica would not be able to run on a normal PC (at the moment atleast)

Since you won't describe this with any more detail, I'll have to take your word for it. However, I refer you to your statement earlier that this AI will "not need 10 super computers, i do not even need 1. A normal PC would do just fine for my AI, but it would need to be modified a little so that it suites our AI a lil better."

Give me those 4 entries (1 kb) and ill have it solve that problem without guessing.

Yes, and it will, believe me it will. It will not get the maze right on the first try. Now that would be godlike. But the door it will solve.

I can't stand discussions where the other person isn't consistent. Let me quote from your previous posts again.

However if it is just an empty room then it would try to do something with the roof but only once till it realises that it cannot do anything with it.

Yeah, you are right. What i meant was that everything appear in text and the roof does not appear so it can't be used. However if it is just an empty room then it would try to do something with the roof but only once till it realises that it cannot do anything with it.

Do you not see that these two pairs of statements conflict? Either it will do it right the first time, or it won't. If it's adaptable as you say, surely it will be able to perform the same whether there's a roof or not.

Yes, it was decided after i first posted here that it will have the goal to get from point A to point B and therefore it has to open the door. We decided this cause otherwise it might try to do something with the apple and the pear first before it uses the key on the door. If it knows that the door is its objective then it will use the key. That still remains

How does it know point B is beyond the door? Couldn't it just as well be hidden behind the wall opposite the door?

I still have a major problem with how you expect an AI with no knowledge base to magically know that the object is a key, that a key is what it needs to open the door, and how to operate that key in the door.

Here is this same question restated, as I posted in my last post:

On top of all of this - particularly if the AI is given no knowledge, why should it even expect that putting the key in the lock and turning it will open the door? Presumably you might say that this is the only option - there's a door, a key, and a keyhole. If I add more complexity to the problem, like adding various objects to the room, not necessarily key-like, why should the AI pick the key instead of the apple? Or the pencil?

yes, identify is detect the object and create a entry about it.
That would allow it to use it (shape like) but not fully. Id like to say some more here but i cannot do that, i do not like competition especially if someone is smarter than me.

You're not understanding me.

1. Recognize there's an object on the floor (distinguish the object from the background).
2. Examine the object (create your 3D model).
3. From the results of the examination, classify the object.

Creating the model is not the same as realizing there's an object there to begin with. Step 1 is what mouse's reply was asking about.

Quite impressive - can you faithfully identify an apple in this setting? How do you know it's not a mutant banana with an edge that happens to look like an apple?

Already been taken care of. That is really the largest problem in this world.
Lets say you have a can of Coca Cola. You put it in your hand and press it as hard as you can, to you it is now a disformed can of cola but to a robot what is it? it would see it as a new object (unless it can see the text "Coca Cola" and understand that it is a disformed cola can. Well, i have a few concepts that i would integrate in highly advanced robots but yes, it can still fail. But the more it know the more accurate)

I understand labeling and the reasons for it. That's not my problem. You can call it a "key" and the AI won't know anything new about it from that label. My problem is that you're recognizing (not identifying) the key for the AI. You're telling it that there is some object sitting on the ground in front of it. This information is not contained in the real world test.

In the real world it has eyes, it will just need the right hardware.

Therefore, you are translating the input.

Yes, in the simulated reality i am translating it, it is the easiest. in the real world i will probably too but let me just point out that i in no way have to.

Because you translate and filter the input in the text simulation, there are marked differences between it and the real world test. Therefore, your results from the text simulation will not indicate feasiblity in a real world test.

Fine, ill make one translated version and one with randomly created strings just for you ;)

I can't stand discussions where the other person isn't consistent. Let me quote from your previous posts again.

My greatest virtue is that i do not see myself as the smartest and the best. I see myself as one that can think of intelligent short cuts. That is how i have done my math, creating an own way of thinking that suites me better everytime that i find something hard. So, i believe that everybody will know all that i know cause some of you in this discussion might even be professional AI people, but the truth is that people don't. I spent 5 hours explaining my AI, i could had done it in 2 hours but i fogot to say one crucial thing cause i took it for granted.

Do you not see that these two pairs of statements conflict? Either it will do it right the first time, or it won't. If it's adaptable as you say, surely it will be able to perform the same whether there's a roof or not.

No, i meant a empty room that nothing intelligent could be done.

How does it know point B is beyond the door? Couldn't it just as well be hidden behind the wall opposite the door?

well, it has been decided to give it the task to get from coordinate 10,10 to 10;10.

I think we've probably gotten as far as we're going to get in this discussion. You're not quite seeing a few of my major points.

In the real world it has eyes, it will just need the right hardware.

I know that it has visual capabilities in the real world test. What I'm saying is that is has to decide what is an object and what is not in the real world. In the text simulation, you decide what is an object.


Yes, in the simulated reality i am translating it, it is the easiest. in the real world i will probably too but let me just point out that i in no way have to.

How will you prove you don't need to if that's what you do?

Fine, ill make one translated version and one with randomly created strings just for you ;)

No, you're not getting it. I don't care what the strings contain. Whatever you call the object doesn't matter. You can't create random descriptive strings, or they won't describe the object, will they? It's these descriptive strings that mark the major difference between text and real world. I'll say it again - in the text simulation, you describe the object for the AI. In the real world, it has to describe it for itself.

No, i meant a empty room that nothing intelligent could be done.

I see. However, this brings up the other point I made. Just because the AI might be able to exit the first room which is empty except for a key and door...what makes you think it'll be able to leave a room with a key, a door, and twenty other objects, in any reasonable amount of time? If it uses brute force, it could try to use each of the 22 objects on each other. If it has no knowledge, why would it decide to pick up the key and go straight to the door?

well, it has been decided to give it the task to get from coordinate 10,10 to 10;10.

Ok.

No, you're not getting it.

No, i know. I was just joking with thee.

I don't care what the strings contain. Whatever you call the object doesn't matter. You can't create random descriptive strings, or they won't describe the object, will they? It's these descriptive strings that mark the major difference between text and real world. I'll say it again - in the text simulation, you describe the object for the AI. In the real world, it has to describe it for itself.

Well, if they were descriptive strings then i could not replace them could i?
You say Apple, i say "Äpple". You say Key, i say "Nyckel". It still means the same thing, just different languages.

I see. However, this brings up the other point I made. Just because the AI might be able to exit the first room which is empty except for a key and door...what makes you think it'll be able to leave a room with a key, a door, and twenty other objects, in any reasonable amount of time? If it uses brute force, it could try to use each of the 22 objects on each other. If it has no knowledge, why would it decide to pick up the key and go straight to the door?

You have gotten that wrong. There will be 10 objects in room one and it will pick the key and use on the door (after it has identified everything cause we tell it to identify everything before interacting in SC-11)

why would it decide to pick up the key and go straight to the door?

Since it has the goal to get from Point A to Point B and the door is the first obsitcle.

I know that it has visual capabilities in the real world test. What I'm saying is that is has to decide what is an object and what is not in the real world. In the text simulation, you decide what is an object.

k, ill be brief.
Yes

I think we've probably gotten as far as we're going to get in this discussion. You're not quite seeing a few of my major points.

k, next reply. Write down you major points cause i think i have answered all that you have asked.

It is of course pre-programmed to some extent. Of course it needs to know how to move. I said that it was Real Artificial Intelligence and not Divine Artificial Intelligence.
It is interesting to note at this point that humans, and many animals, do learn how to move. We are born with an instinctive urge to learn how to walk as quickly as we can, but have no pre-programmed knowledge of knowing how exactly we should do that.

That i cannot say. It is not full back-propagating. Maybe i should had said that it uses a neural nets concept instead. This only confusses it i see.
How can it be not full back-propagating? At want point does back-propagation stop?

The professional AI developer in my team has already called it "Neural Genetic Natural Language" He, just as me think that it is the core of AI, "mixing" all the other AI and combined creating real AI. (Not all concepts, ill not say which)
Mixing entirely different concepts of AI is quite an effort. Simply tossing them together obviously is not going to work. E.g. GAs producing efficient NNs? It is not going to happen without throwing either an amazing set of hardware at it (which you suggested you didn't need) or a really crafty and resource friendly method of coding a NN, and all variables that can be associated with it (topology, weight modification algorithms, etc.), on a gene set. Of course, I'm quietly ignoring the ridiculous amount of time it will take to tweak such a GA for getting out local optima and such. I'd be afraid of the amount of other problems I will encounter if I start to think longer than two minutes about it.

No, reality. It creates the mimick, i never said that it would use the same words as we do.
How can you create something you are going the mimick? I'll try again: humans learn language and speech from others. An individual human baby e.g. does not learn a language, as we know it, simply by being in an empty room with a key and a door. The child will perhaps associate certain labels to the room, the key, walls, and what ever is there, but aside from that it will be deprived of something as complex and usefull as English or any other language which took centuries, if not millennia, to develop.

However if it was to have audio input then it could mimick the speakers of a conversation and use that data in many fields.
You are making a quantum leap here. Understanding spoken language is very difficult. It takes humans years to fully master it, and with good reason. Take the sound of the word "to" e.g. Based on context it can be associated with a verb "to walk", or, as a superlative (? not sure if this is the correct term) as in "too much", or as a number as in "two keys". If even humans, equipped with an impressive bunch of wetware, need long exposure to a language to figure this stuff out, how do you propose your AI can understand language in a reasonable amount of time with common hardware?

Yes, the only disadvantage is time. I press a button in room one, how the hell could i possibly know that it opened the door in room 5? I dont think there is any way actually except the concept of Origin Tracing, one of the earliy features of my design.
My problem is not with an action leading to an event it can not see or hear. My problem is this: if you associate an action with an event, how do you know which action you are going to couple with which event. While I press the buttons of this keyboard, I'm hearing music, a chopper flying at some distance, people chatting in a corridor and occassionly a phone ringing. I am seeing text appearing in a form, but I also see plants waving in the wind, I see cars driving by. While I'm hitting those buttons, I'm also performing an extensive array of other actions: breathing, regulating digestion, pumping blood around, and many many other issues I am luckily not consciously aware of. In short, while my finger hits a button, a lot of events are happening which I could associate with hitting the button, or perhaps any of the other actions I was performing at the same time. How does your AI decides which events can be associated with which action, and which not? Mind you, to discover this by just trial and error would be inmensly resource expensive, given the amount of events and actions in a real world situation.

k, what is specific of walls? If you can only see text? Not colors, not material.
To answer that, I'd need to know how you textually describe the wall to your AI.

Since it has the goal to get from Point A to Point B and the door is the first obsitcle.
Why? Isn't the wall the most probable first obstruction? Or is point B directly behind the door?

k, next reply. Write down you major points cause i think i have answered all that you have asked.

You've quoted most of the points, but you usually only address certain issues related to the points. Here they are:

1. Key differences between a text simulation and a real world test mean a success in a text simulation does not imply a real world application will succeed.

a. The major difference is with object detection. In a text simulation, you detect the objects for the AI, which it will then proceed to identify based on your descriptions. In a real world test, it must detect them on its own based on video input.

b. A related difference is with object identification. In a text simulation, you describe the objects in some manner the AI can understand. In a real world test, it must determine the objects' characteristics based on video input.

2. An AI with no initial knowledge base is expected to determine that the key is the object it needs to open the door.

a. There are ten objects in the room for which the AI has no predetermined classification. There is absolutely no reason the AI will pick the key first, except by chance.

b. Related is the expectation that the AI will grasp the concept of a key. That is, it will understand what a key is - what it looks like, what it's used for. If the AI has no initial knowledge base, where does this knowledge come from? If you have no knowledge of keys or doors and I handed you a key, would you know what it's used for? No.

3. Practical implementation of the knowledge base

a. Since you won't give more information on your database, I'm unable to pursue this any further. However, I'm naturally skeptical that you can contain and search the amount of knowledge required to produce behavior equivalent to that of a human.


These are my major points. Incidentally, I also have a problem with how you define your goal. In a text simulation, telling the robot to move from coordinate 0,0 to 10,0 might be feasible since the simulation could provide the robot with information on its location. However, in the real world, there is often no such provider of information, except perhaps GPS. How will the robot know that coordinate 10,0 is behind the door? Because that's 10 coordinates in front of it? What if you turned it 90 degrees before you started the AI?

k, ill answer malkiri's questions first.

Ill try to address your problems as well as i can.

The major difference is with object detection. In a text simulation, you detect the objects for the AI, which it will then proceed to identify based on your descriptions. In a real world test, it must detect them on its own based on video input.

Yes, that is right.

A related difference is with object identification. In a text simulation, you describe the objects in some manner the AI can understand. In a real world test, it must determine the objects' characteristics based on video input.

Yes, id say that is correct too.

So, how could it be implemented in the real world? Why does it has too be any difference? No matter which "world" you are in there still remains one constant (However, i cannot tell thee, for all i know you might be working for the "enemy" (all are my enemies till the opposite is proved))
I guess that it is this you would want me to address, right? Contact me and if i believe that i can trust you then ill fill you in on my secret.

There are ten objects in the room for which the AI has no predetermined classification. There is absolutely no reason the AI will pick the key first, except by chance.

that is what you would think, yes. But i try my best to give (whoever) the WOW feeling. Especially important if we try to implement it in millitary applications as we planned to do in the first place.

Related is the expectation that the AI will grasp the concept of a key. That is, it will understand what a key is - what it looks like, what it's used for. If the AI has no initial knowledge base, where does this knowledge come from? If you have no knowledge of keys or doors and I handed you a key, would you know what it's used for? No.

No, i would only be able to guess. So you would say that it is super-human intelligence then? Well, it is conforting to know that my AI is better than humans in some fields atleast since i know that it will not be as good as humans in other.

Since you won't give more information on your database, I'm unable to pursue this any further. However, I'm naturally skeptical that you can contain and search the amount of knowledge required to produce behavior equivalent to that of a human.

Yes, and i like people who are sceptical. It just provides me with a greater challange.
But i would not want people who merely say Yes, the people who sometimes say No and ? are sometimes needed. (atleast on this project so that i can solve all the problems that might occur)

These are my major points. Incidentally, I also have a problem with how you define your goal. In a text simulation, telling the robot to move from coordinate 0,0 to 10,0 might be feasible since the simulation could provide the robot with information on its location. However, in the real world, there is often no such provider of information, except perhaps GPS. How will the robot know that coordinate 10,0 is behind the door? Because that's 10 coordinates in front of it? What if you turned it 90 degrees before you started the AI?

Well, that is a real question.
GPS? maybe in a human replica but no, not otherwise.
No, if it knows the concept of movement and knows that point B is 100 cm in that direction then it could update point B's location accordingly to its movement. It will also memorize the maze.
Bet that i did not address it like you wanted me to niether?

Malikiri, mail me instead.

k, mouse

Why? Isn't the wall the most probable first obstruction? Or is point B directly behind the door?

sure, they are equal. but it cannot interact with the wall. and if it tries then it would fail.

It is interesting to note at this point that humans, and many animals, do learn how to move. We are born with an instinctive urge to learn how to walk as quickly as we can, but have no pre-programmed knowledge of knowing how exactly we should do that.

Are you not forgetting that newborn babies can swim? an abbility they lose at a later point. They innherit the info in their genes on how to walk but it might not be connected fully.

How can it be not full back-propagating? At want point does back-propagation stop?

Mixing entirely different concepts of AI is quite an effort. Simply tossing them together obviously is not going to work. E.g. GAs producing efficient NNs? It is not going to happen without throwing either an amazing set of hardware at it (which you suggested you didn't need) or a really crafty and resource friendly method of coding a NN, and all variables that can be associated with it (topology, weight modification algorithms, etc.), on a gene set. Of course, I'm quietly ignoring the ridiculous amount of time it will take to tweak such a GA for getting out local optima and such. I'd be afraid of the amount of other problems I will encounter if I start to think longer than two minutes about it.

No, you are not getting me. I have a totally unique AI but it is similar to NN, Natural Language, Pattern Recognition. However it can provide the same results as Expert Systems, GA and almost everything you can imagine, it is human. That is all i can say about it.

How can you create something you are going the mimick? I'll try again: humans learn language and speech from others. An individual human baby e.g. does not learn a language, as we know it, simply by being in an empty room with a key and a door. The child will perhaps associate certain labels to the room, the key, walls, and what ever is there, but aside from that it will be deprived of something as complex and usefull as English or any other language which took centuries, if not millennia, to develop.

Yes, how can i? why does it has to learn an language as we learn it when it is not human itself?
k, patterns alter the pattern of thinking accordingly to the environment.

You are making a quantum leap here. Understanding spoken language is very difficult. It takes humans years to fully master it, and with good reason. Take the sound of the word "to" e.g. Based on context it can be associated with a verb "to walk", or, as a superlative (? not sure if this is the correct term) as in "too much", or as a number as in "two keys". If even humans, equipped with an impressive bunch of wetware, need long exposure to a language to figure this stuff out, how do you propose your AI can understand language in a reasonable amount of time with common hardware?

Not really, i just use my universal algorithm in reverse.
This is however an interesting topic and some problems will probably occur but we will only use this later on in (probably) military application so it is an topic for discussion then. But no problems that i cannot solve.

My problem is not with an action leading to an event it can not see or hear. My problem is this: if you associate an action with an event, how do you know which action you are going to couple with which event. While I press the buttons of this keyboard, I'm hearing music, a chopper flying at some distance, people chatting in a corridor and occassionly a phone ringing. I am seeing text appearing in a form, but I also see plants waving in the wind, I see cars driving by. While I'm hitting those buttons, I'm also performing an extensive array of other actions: breathing, regulating digestion, pumping blood around, and many many other issues I am luckily not consciously aware of. In short, while my finger hits a button, a lot of events are happening which I could associate with hitting the button, or perhaps any of the other actions I was performing at the same time. How does your AI decides which events can be associated with which action, and which not? Mind you, to discover this by just trial and error would be inmensly resource expensive, given the amount of events and actions in a real world situation.

That is a great question.
Well, if your first interaction method on a computer (without any errors) are using the mouse and the keyboard then why should you need more empirical data that the keyboard and mouse is doing what you see on the screen? I mean if you already will try to interact by using the buttons then "why change it if it aint broken"?
In a full human replica it would to the full extent as you would imagine. That is all i can say.

To answer that, I'd need to know how you textually describe the wall to your AI.

Dont actually describe the walls since it has no function. But i would describe it "Wall" and that is all that it has to know.
To open the door it has to know "Door", "Key" and that they are objects (object=something that it can see or touch and possibly interact with.) It also needs to know "Unlocked" and "Locked" and that they are the 2 statuses of the door.
The words does not matter since it will be able to use it in the same manner no matter name but i think you all get that by now.


Anything more?

Can we lock this thread? Baal has moved from being ambitious, to lying, to being delusional. Now he's just 'playing along'.

Hell, statements like "Not really, i just use my universal algorithm in reverse" don't deserve to be in a science forum. If I came on here and said I was going to make an AI by waving my magic wand backwards I'd expect it to be closed. Baal is just using pseudo-scientific terms... most of which are used incorrectly.

sure, i do not mind.

ambitious, to lying, to being delusional
Ambitious i can agree with, Lying that i have never done. Delusional, the only thing that i am delusional is when i am tired and see things moving that isn't really there.
Playing Along, yes. Thought that would be a better attitude.

Let me just explain, why it sounds so strange.
I do not have experience in other AI, i came up with my AI all alone. Do you know what CBR is? Well think the CBR technology 3 times as advanced combining some features from NN then you have my AI.
How can you combine some features from NN with CBR? well, if that is what you are interested in hearing then maybe i came to the wrong place in search of good programmers.
"If the facts don't fit the theory, then change the facts"
That is all i have to say.

sure, they are equal. but it cannot interact with the wall. and if it tries then it would fail.
But the point is that it tries. You have no way of saying in advance, if the AI has no initial knowledge about doors and walls, what it would go for first.

I have a totally unique AI but it is similar to NN, Natural Language, Pattern Recognition.
"I have a totally unique computer, it is similar to Object Oriented programming, C++ and web browsing", is a statement equally unintelligible to your statement.

However it can provide the same results as Expert Systems, GA and almost everything you can imagine, it is human. That is all i can say about it.
And based on that utterly unintelligible statement, you claim to have the holy grail of AI.

why does it has to learn an language as we learn it when it is not human itself?
The fact that it needs a bunch of humans and much time to develop a language, does not make it "not human". Learning a language is one of the things that we obviously are capable of, and is often regarded as one of the signs of intelligence.

Not really, i just use my universal algorithm in reverse.
I agree with Persol on this. Do you really mean this seriously, or are you just joking around and having fun here? In the first case, get help and a proper education. In the second case, get lost.

This is however an interesting topic and some problems will probably occur but we will only use this later on in (probably) military application so it is an topic for discussion then. But no problems that i cannot solve.
How do you know that in advance, if you do not know what problems you would encounter?

That is a great question.
No, it was quite a fundamental one. The fact that you have to think about it proofs that your project is in an infant state, if it exists in the first place.

Well, if your first interaction method on a computer (without any errors) are using the mouse and the keyboard then why should you need more empirical data that the keyboard and mouse is doing what you see on the screen?
That's knowledge. Somebody told me how computers work. Nobody has told your AI how keys work.

Dont actually describe the walls since it has no function. But i would describe it "Wall" and that is all that it has to know.
But you'd have to code which actions are allowed on the wall, and which not.

To open the door it has to know "Door", "Key" and that they are objects (object=something that it can see or touch and possibly interact with.) It also needs to know "Unlocked" and "Locked" and that they are the 2 statuses of the door.
That is incredibly simplistic. As Malkiri tried to point out to you, it is a far cry from a real world simulation.

Anything more?
Lots, but I'm afraid there is little use. You do not answer the questions in concrete terms, but rather by:

concatenating AI terms in random order,
by an unjust simplification of the problem presented until you can easily wave it away,
by ignoring the problem presented altogether,
or by stating that you can not tell more out of fear of competition.

I suppose you are indeed in the wrong place if you think you can enlist programmers with such a shaky argument.

I agree with Persol on this. Do you really mean this seriously, or are you just joking around and having fun here? In the first case, get help and a proper education. In the second case, get lost.

Actually, when using data you run the algorithm as usual. When learning from either hearing or seeing you run it roughly in reverse.
Pretty obvious that the opposite from leanrning is usage of learned data.

"I have a totally unique computer, it is similar to Object Oriented programming, C++ and web browsing", is a statement equally unintelligible to your statement.

:)
Well, you have a computer that can do both of those things but somehow you do not see your computer as a holy grail, or do you?

No, it was quite a fundamental one. The fact that you have to think about it proofs that your project is in an infant state, if it exists in the first place.

Just because i say "Great question" does not mean that i have to think about it. It just means that it is one of the questions that actually touch my AI.


I suppose you are indeed in the wrong place if you think you can enlist programmers with such a shaky argument.

well, as you wish. if you actually want to hear anything about the AI then just mail me.
I will stick to my professionals instead.

How do you plan on this identifying object using video?

i deleted my reply to that question. If you want the answer then mail me instead.

Perhaps a question you can answer.

How do you plan to solve a dozen on more complex problems that the field of AI has been completey unable to solve, do it with a normal computer, and do it with random programmers? If you don't want to answer, then fine... but you aren't going to get many volunteers. Your idea is very 'pie in the sky' and unless you can back it up I don't see people wasting thier time on it.

let me first ask you,

Why would i post something that was not true?
Secondly,why would i not post more unless it was true?
Respect can obly be achieved by having actuall proof.
So?

I'm not saying you don't believe it is true. I'm saying that you haven't demonstrated that you actually understand the difficulties and barriers to what you propose.

You are promising the 'holy grail' of AI... yet seem to not see the practicalities that stand in the way. These include:
shape identification via video
3d modeling from shape identification
object pattern matching
how an AI 'without knowledge' picks the right solution on the first try
'good' computer simulation (IE: collision detection and object interaction)
storage method of knowledge learned
method of converting input to data
method of logic (A is B so C is D)
processing speed (how do you do this on a standard computer)
-expontential growth problem. more knowledge leads to slower speed.
language processing

etc, etc, etc.

There are many many many things that your AI has to do, each of which is basically a field in itself. You don't seem to recognize the scale of what you are doing.

If Baal and his team are that dedicated to their cause, then they will most certain do what is necessary to achieve their goal.

Just because there isn't a "HOWTO" or "DUMMIES Guide" to building an A.I. being posted on the board doesn't mean that their intensions of building one aren't genuine.

I would suggest Baal that you might think along the lines of doing this perhaps as a larger project, which you might suggest doesn't have some of the monetary benefits.

Admittedly you would have to e-mail/message me via Hotmail/MSN to get an understanding of how I mean.

If Baal and his team are that dedicated to their cause, then they will most certain do what is necessary to achieve their goal.

Just because there isn't a "HOWTO" or "DUMMIES Guide" to building an A.I. being posted on the board doesn't mean that their intensions of building one aren't genuine.I'm not saying his interest isn't genuine, but that his saying he 'has the answer' isn't genuine. Hard work alone doesn't provide results when you don't they don't even know the scale of what it is they are trying to accomplish.

Stryder, You have been nothing but nice to me. I thank thee and unlike other forums SciForums is a place that i would like to be a member of.

I accept your offer to chat with thee.

Persol, ill mail thee instead.

So, how could it be implemented in the real world?

This is exactly the problem.

Why does it has too be any difference?

Because there's a difference between the real world and text.

that is what you would think, yes. But i try my best to give (whoever) the WOW feeling. Especially important if we try to implement it in millitary applications as we planned to do in the first place.

Again, you don't get it. I'm not saying your AI in particular will not be able to do it. I'm saying any intelligence whatsoever cannot do this on the first try unless it has prior knowledge.

No, i would only be able to guess. So you would say that it is super-human intelligence then? Well, it is conforting to know that my AI is better than humans in some fields atleast since i know that it will not be as good as humans in other.

That's not at all what I was saying. I'm saying your AI will not be able to do it.


Well, that is a real question.
GPS? maybe in a human replica but no, not otherwise.
No, if it knows the concept of movement and knows that point B is 100 cm in that direction then it could update point B's location accordingly to its movement. It will also memorize the maze.
Bet that i did not address it like you wanted me to niether?

No, you didn't. I don't care about movement, although that is a separate issue. I'm talking about simple orientation in relation to the goal.

I don't buy into your idea to warrant spending any more time with it. You're not really answering my questions sufficiently, either because of a language barrier or simple avoidiance because of faults in your concept, not to mention your reluctance to share certain parts. Good luck with your project. Negative as it may sound, I neither believe nor hope you will succeed; the former because I think you're shooting far too high, and the latter because I believe your secrecy will not serve the field of AI, though I understand that different people have different motivations.

Persol, ill mail thee instead.Please PM me. (Basically email within this forum.)

My impression is that you don't grasp the size of what you are trying to do, but I look forward to a few more details. I know basic, C, pascal, and can probably learn something else if you program in that instead.

Thanks.

Ohh the pain the pain... Come on people is surprises me that you don't understand the meaning of AI...
Things to consider.
First AI means human like intelligence. It does not mean intelligence made by humans.
Second. Intelligence is inherent in all turing machines.
Anyone who has ever created an "if then else" statement will have created an intelligent machine.
Baal Zebul. You should stop using the acronym AI because your project is not an AI. You will save your self a lot of grief when selling the project.

malkiri, How to create a fully functional robot (in the real world), that is your question.
Well, i have many conservative people on my crew but all of them are negative in an way that the concept profits from.
If it was not for Vincent who said "No, it lacks [something]" then i would not have come up with everything that i have now. But your just saying "No" and not what it lacks.

Let's just put it like this, i have solved the problem with implementing it into our world. How? Well that will remain secret so that the AI community does not profit from it.

Artificial Intelligence is using pre-programmed data "intelligently"
Whilst i rather say that Real Artificial Intelligence features obtaining that data.

Id say the prmary reason is my inability to trust.
If you ever feel that you know better swedish than i know english then send me a PM. :)
If this was one month ago then i might had said more but one of my members left and stole my AI concept, unfortunatly he is not as clever as me so he is trying to solve the real world integration with Pixel-Based pattern recognition (analysing a image pixel for pixel). So if you ever want somebody to harass then i could give you his hotmail because all your critisism fits what he is doing.

Update:

Need someone who is extremly good at extremly advanced math. We are working on a hard problem for real world integration. We have the concept complete but we need the math correctly in order to be able to code it.

I design strong AI and have a web-site dedicated to that problem:
<a href="http://www.dennisgorelik.com/ai/overview.htm">http://www.dennisgorelik.com/ai/</a>

Leda, I enjoy your questions. They are strait to the point.

What data structures does it (AI) use to hold knowledge?

The core memory consists of two database tables.
1) <a href="http://www.dennisgorelik.com/ai/ConceptTable.htm">Concept table</a>.
2) <a href="http://www.dennisgorelik.com/ai/CauseEffectRelationTable.htm">Cause-effect relation table</a>.

So, every concept can be related to any other concept.
Every relation has a weight.
The higher the weight is, the stronger cause-effect relation between the concepts is.

Just give you a hint about the size of the tables.
For instance, Concept table could have 10 million records (concepts)
Relation table could have 200 million records (cause-effect relations).

Any questions are very welcome! :-)

How are you planning on representing objects?

Every object in the real world can be represented by one concept in AI memory. Actually not just one concept, but concept + many relations to other concepts (and other concepts also relate to the next set of concepts).

<a href="http://www.dennisgorelik.com/ai/Concept.htm">Concept</a> alone holds very small amount of information, but together with references and other concepts anything can be described.

Some concepts in AI memory are connected to <a href="http://www.dennisgorelik.com/ai/PeripheralDevices.htm">peripheral devices</a>.
That solves "Grounding" problem.

Blindman:
Ohh the pain the pain... Come on people is surprises me that you don't understand the meaning of AI...
Things to consider.
First AI means human like intelligence. It does not mean intelligence made by humans.

Er, artificial means constructed, not "human like". Since other animals do not make AIs, we humans are generally forced to try our hands at it if we ever want to see one.

Second. Intelligence is inherent in all turing machines.

Here is a Turing machine:

1. Halt

Is that intelligent? Well, no... it's about as intelligent as a grape. (Probably less - a Turing simulation of a grape would probably be much more complicated.)

Anyone who has ever created an "if then else" statement will have created an intelligent machine.

Eh... that's not really what anyone's talking about. Baal's talking about a relational engine that anticipates a series of actions required to achieve an arbitrary goal in a simulated world. "If then else" isn't going to cut it here.

Baal: Your AI simulations aren't going to tell you anything about building an AI in the real world. Why bother? Why not just start with real-world prototyping? You'll probably learn a lot faster this way.

Dennis Gorelik (if that is your real name)

First, whereas you may be a real person, your AI website is crap - just a series of definitions of words that all refer to one another. I skimmed your documentation and this is all it contained - no suggestions of how your AI is made or what it does, no description of AI theory that couldn't be gotten out of a dictionary.

To be honest, your page looks like it was assembled by a bot. And... not an impressive bot either.

So, just to assuage my fears that you don't know anything, maybe you could answer me this...

Just give you a hint about the size of the tables.
For instance, Concept table could have 10 million records (concepts)
Relation table could have 200 million records (cause-effect relations).

So basically, you have a series of nodes, with paths between them that have certain weights. The trick of determining a germane relation to the series of inputs the robot just received, then, will be to find the collections of nodes that relate to the input, that have the highest total weight to their paths.

I believe in school we called this the Traveling Salesperson Problem. The main problem with this is that the complexity of the problem is N!, which grows very quickly as N grows.

This means that, in a set of - say - twenty fully interconnected nodes, finding the optimum solution by direct comparison means testing

2432902008176640000

possible paths.

Assuming that you can test one of these every nanosecond (which would be pretty fast), that means the calculation of these solutions would take you 2432902008.17664 seconds

= 40548366.802944 minutes

= 675806.1133824 hours

= 28158.5880576 days

= 77.1468166 years.

So, expect to die of old age waiting for your robot to process its first few inputs.

Incidentally, if there were 21 fully interconnected objects, expect this problem to take (77 * 21) = 1617 years to solve. 22? (1617 * 22) = 35574 years.

So, how do you intend to breeze past the mathematical problem (solving NP-complete problems within a reasonable time) that the entire world is still struggling with?

Take your time...

(Baal... do you have a solution? The fast expansion of relational systems is one of the biggest problems in AI, so you'd make a lot of people really happy if you came up with an answer.)

Well, actually i have two new documents now.
CES Artificial General Intelligence and Machine Vision (An introduction to SC-11 technology)
and
CES Intelligent Neural Networks

Well, i have changed the structure to a Neural Nets. But it is nothing like other Neural Nets, but the process flow has always been similar to Neural Nets.

Btw, my real name is Rikard Svensson

Here is a copy from our website:

SC-11 - Number1
Number1 will simulate a high-poly 3D forest with lots of trees, a blue sky and some clouds. The simulation will have a dynamic terrain and situated on it will be a Woodcutter, swining an axe at a tree. The Number1 simulation will use the latest technology (EDS, WFS and SRS) developed by Rikard Svensson in order to be able to mimic human movement without any prior knowledge by just viewing a visual stream. The AI used in Number1 is beyond the state of the Art AI by many years.


SC-11 - Number2
Number2 will simulate a short sequence of a bouncing ball. By having this information the Artificial Lifeform (also known as ALF) will use Origin Tracing (or OT) in order to calculate the origin of the bouncing ball and its presumed final destination. Origin Tracing is designed so that by just seing a short sequence of the bouncing ball it can still determine the gravity of the world (will be set to 9.81 m2/s) and it can also determine various features of the ball such as outer hull strength, texture, weight (if gravity is known) and much more.

By designing a more complex version of the Number2 simulation it is also possible to detrmine Usage Area of the ball by having a simple 2D visual stream.


SC-11 - Number3
Number3 will be the last real computer simulation in the SC-11 series. Number3 will simulate a highly trafficed street and the ALF's task is to cross the road. Number3 might sound trivial compaired to Number1 but actually Number3 will require more usage of the latest CES technology.

Number3 will have a limited window of opportunity to cross the road and it will have to use all it's technologies to seize the moment. The ALF can only look right, left or forward instead of having a overview of every action in the world at once. There is no other AI that can solve this without a optimized solution for that particular problem whilst the CES AI is universal.

Number3 will cross the road on the first try, there is no trail and error neccesary. If it would fail then it is because some mathematical equations are wrong. The CES AI cannot fail since it is not controlled by code, it is controlled by environement and knowledge.


Given that it can solve all those problems then it is ready for real-world integration. We will make a solution to Darpa's Grand Challange.
And so that you do not think like Dennis:
All the simulations will use the same universal solution and not a manufactured solution for every simulation. However SRS and RSS are totally different things and the ALF can use one (not both).
What EDS, WFS, SRS and RSS means i will not write here.
And i forgot to say, these samples focus mostly on machine vision rather than the AI, that we will have different samples for and i have not finished designing them all yet.

BigBlueHead, would you be interested in a chat this weekend? (i will be online 24 hours every day of the weekend)
My MSN messenger is, da_rikard@hotmail.com
I also have ICQ but not installed on this computer but if i really have to ill download it.

Same goes for you Stryder, if you are interested in a chat then please add me or tell me your ICQ or something (i do not have AIM)

/Rikard Svensson - CEO, The CES Project

My pet project AI

The programming language is immaterial.
The brain uses mathematics, but is of a sort that is not known by humans.

It is the mathematics of symbols, but not of numbers... I assure you there is a difference.

This is why we write in letters or speak in sounds.... all mathematics.

I would love to access a C programmer... sigh

Zarkov,

The programming language is immaterial.
No, it is not. Some programming languages are more suited for one type of problem than others.

The brain uses mathematics, but is of a sort that is not known by humans.

It is the mathematics of symbols, but not of numbers... I assure you there is a difference.

You seem to be referring to predicate logic (http://www.cs.odu.edu/~toida/nerzic/content/logic/pred_logic/intr_to_pred_logic.html). It is well understood. However, as a method for creating AI it is not adequate.

This is why we write in letters or speak in sounds.... all mathematics.
We write in letters or symbols because it is a convenient way of communicating. Not because it is "all mathematics".

I would love to access a C programmer...
Why not learn it yourself (http://www.physics.drexel.edu/courses/Comp_Phys/General/C_basics/c_tutorial.html)? Forcing yourself to think in an abstract way and being pushed in a logical framework wouldn't do you any harm.

>> Why not learn it yourself

I have written C a long time ago, the modern approach with graphics as per java, is a little daunting.

But yes my problem is just doing it amongst other things.... an experienced programmer can outperform me at the speed of light.

Thanks for the link, I do intend to devote time to this project, but not for general consumption... err maybe.... I have developed a method of setting a brain up, and only have internal links with limited knowledge, so it becomes impossible to determine the program sequence as it can be deleted ( bit like rote knowledge ), but it would be indistinguishable from a human, except it could not learn.... but it could dish out intelligence and a huge information base.
I would retain the learning brain, and clone rote copies......

I do not think the world is worth it though....

>> No, it is not.

Well I have a model written in basic... no language is immaterial

>> You seem to be referring to predicate logic.

no, I said it is all mathematics.... what else do you think can be in the brain, even images are a mathematical matrix... really all electrical signals that are process by the rules of mathematics, but 1+1 does not equal 2 as numbers are but an image of real mathematics.

Thanks for the tutorial mouse, I will start the project in C.

Is a C compiler available to download ?

That is why my project is called The CES Project, CES is short for Cogito Ergo Sum ("I think therefore I am")
Descartes said that everything in this world could be explained by mathematics but just like Zarkov mine is based on language. He might say that it is math now but in about 10 months he will realise that it aint.

Zarkov,
what else do you think can be in the brain, even images are a mathematical matrix... really all electrical signals that are process by the rules of mathematics, but 1+1 does not equal 2 as numbers are but an image of real mathematics.
Knowing that neurons send electrical pulses to eachother doesn't bring you significantly closer to building an AI. You are still left with an enormous amount of problems to solve. In this thread a number of them were discussed.

With regard to C-compilers: if you are running a Linux distribution, most likely gcc is already installed somewhere. For Windows I do not know a good freeware C compiler. You can use Microsoft's C# though, which is most definitely not C but an interesting language nonetheless. It comes with the .Net framework SDK. You only have to sell your soul to download one of those.

http://msdn.microsoft.com/visualc/vctoolkit2003/

Alternatively, you can use Cygwin to use gcc.

Also, there's djgpp: http://www.delorie.com/djgpp/

Thanks I will boot up a linux machine for the job.

Maybe if I have some C problems (syntax etc) someone here may be able to help ..???? yes

>> He might say that it is math now but in about 10 months he will realise that it aint.

I assure you it is all math, the math of symbols (language)... the most powerful for of math, and it makes one really wonder how Darwinian evolution could come up with such concepts....... impossible IMO.

Er, artificial means constructed, not "human like". Since other animals do not make AIs, we humans are generally forced to try our hands at it if we ever want to see one.

Artificial: imitating nature

“artificial intelligence
noun {U} (ABBREVIATION AI)
the study of how to produce machines that have some of the qualities that the human mind has, such as the ability to understand language, recognize pictures, solve problems and learn”

Here is a Turing machine:

1. Halt

Is that intelligent? Well, no... it's about as intelligent as a grape. (Probably less - a Turing simulation of a grape would probably be much more complicated.)

Because it is capable of being intelligent does not require it to do intelligent things.

Eh... that's not really what anyone's talking about. Baal's talking about a relational engine that anticipates a series of actions required to achieve an arbitrary goal in a simulated world. "If then else" isn't going to cut it here.

I would assume that Baal’s relational engine is coded in some machine-readable form. Every language I used to code in is converted into assembly or interpreted by assembly level code. Without a conditional jump you can’t do anything intelligent. If his code did not have a single for, while, case, or if, I would be totally awe struck. So a single conditional jump is dumb but a lot is not? Saying “if then’s are not going to cut it” is saying that intelligence is not achievable on turning machines..(Modern computers).

There are some that question the ability of the turning machines to produce intelligence, if so, Baal is on the wrong trail and should stop pronouncing the intelligent algorithm.

The debate is about whether Baal’s engine is intelligent. I say yes. Confusion with the AI acronym is the core conjecture. I vote that AI represents Human like intelligence and that we except that computers can produce intelligent behavior outside the human sphere

Blindman:

Saying “if then’s are not going to cut it” is saying that intelligence is not achievable on turning machines..(Modern computers).

It may be that they are not, although I wouldn't claim that; what I'm saying is that AI will probably be several levels of emergent property away from the level of Turing computability... hence, trying to use the metaphor of Turing machines to describe an AI may be like trying to describe your trip to Nebraska in terms of the series of chemical reactions involved.

The ability to evaluate conditionals does not denote intelligence, only deterministic behaviour. I think your definition of AI is very restrictive, considering how loosely you define intelligence.

thank you Blindman (i think),

I say that AI is the ability to use pre-programmed data whilst real intelligence is the ability to obtain that data and use it. I would prefer to call it Artificial Intelligence rather than Real Artificial Intelligence but too many have tried and failed for that even being possible.

The first editions of the code contained almost merely If, Then, Else but now a days it does not. It contains a self-created if, else, then created by the machine accordingly to percentage data in the knowledge files. So basically it creates it's own If, Then, Else but i am not going to say that the code does not contain any If, Then, Else's cause that would be a lie.

About the Turing Test, actually id prefer a test which would show that my AI would beat up to 90% of all humans when it comes to problem solving even with lesser data than those humans. My AI is built for problem solving and real world integration, not performing IQ tests which are written to fool the algorithm.

Baal: the degree to which arrangements of information can produce pathological results in your intelligence may be highly relevant.

Certainly, human senses have certain pathological cases where their processing of information is somewhat ambiguous.

Does your system work on a series of weighted averages? I assume it's a hierarchical object sorting system?

Well, it takes information from environment (what object is closest and sees if an stronger object is worth the extra meters/miles it has to walk)
It takes information from empirical data (percentage and combines it with the closest object calculation)
Then it has one more object used in the calculation but i cannot disclose that.

But this is like 10% of the AI concept. If you say that it is highly relevant then the rest would stun you.
Anyone who has got time and are interested in helping us then please mail me. I got the crew that i need but they have not got the time that i need so i need more manpower.

So a closer object is more heavily weighted than a further object in its estimation of importance?

Does it adopt multiple goals at the same time? I suppose it would have to, to be able to get a key and then go to a door to use the key... I guess this is the question I should ask.

Robot is in a room with 8 keys and a locked door. The 8 keys are in different locations in the room. What actions does Robot take, and in what order?

Robot is in a room with 8 keys and a locked door. The 8 keys are in different locations in the room. What actions does Robot take, and in what order?

that is a good question.
Lets say that it's task is to open the door (which is locked) then that would override it's built in will to identify every object algo.
Well 8 keys, 1 right. All keys do the same thing and if it can't see shape then it cannot solve it. Trial and Error.
It could make it on the first try, and it could make it on the last. I said i was re-creating humanity, not custom building a god. No human could do it better, i hope you agree on that.
The robot would identifiy every object (if tasked) or merely try the one shortest in range (if differently tasked).

So a closer object is more heavily weighted than a further object in its estimation of importance?

Yeah, i implemented this as a sub-system for The Republic. Basically so that the ALF would know where the Axe was and where the Saw was since both could chop or saw down a tree so that it could think about which object was the closest and how it could change it's path in order to do it as fast as possible. It was of course told to do everything as fast as possible because the more time it had a day the more new things it could learn and the more it learned, the faster it could do new things. :)

Does it adopt multiple goals at the same time? I suppose it would have to, to be able to get a key and then go to a door to use the key... I guess this is the question I should ask.

Yes, in the "future model" of the AI Net it would have percentages coming with the tasks so it would perform the important tasks first but that is not neccesary and i came up with it whilst thinking of military applications for my AI.
There are two types of tasks. Goals given by humans (orders) and encountered problems in the physical world whilst interacting or observing.
Note that humans don't have to tell it anything and with the latest technology i designed it is possible to initate a robot into this world without any knowledge as long as it can see people move and interact.

must read my post. The turning is "Turing machine.." Opp’s

You can go back and edit old posts - but I think everyone knew what you were talking about anyway, so no biggy.

k,

I need one more programmer

I have two teams and i need one more. Not a team leader.

Funny, I volunteered if you could explain a little more, but you decided not to.

actually you are still on the list of those who i can send information too.

do you have any chat application?

I am a .Net programmer. Been programming for many years in various languages. Am I to believe that you have created an algorithm to simulate life choices?

Blindman:
First, whereas you may be a real person, your AI website is crap - just a series of definitions of words that all refer to one another. I skimmed your documentation and this is all it contained - no suggestions of how your AI is made or what it does, no description of AI theory that couldn't be gotten out of a dictionary.



Ok, tell me in which AI-theory book could I read about importance of <a href="http://www.dennisgorelik.com/ai/Forgetting.htm">forgetting</a>?


What about an idea to use <a href="http://www.dennisgorelik.com/ai/RDBMS.htm">database</a> as the storage for the Neural Net?
Have you ever read about this idea?

What about goal-oriented system?
The idea that several goals (<a href="http://www.dennisgorelik.com/ai/HardcodedGoals.htm">Hardcoded goals</a> or Super goals) have to be predefined during design & development of the AI system. Then the system has to create its own set of goals (millions of <a href="">softcoded goals</a>). Softcoded goals are based on the hardcoded goals. I name it <a href="http://www.dennisgorelik.com/ai/MotivationSystem.htm">Motivation system</a>
Where have you read about this idea?

Blindman:
Dennis Gorelik (if that is your real name)


Yes, it is my real name.
I wonder - why do you use nickname?
Why not real name?

Blindman:
First, ... your AI website is crap ...

Unfortunately you are right.
My goal was to explain my ideas to other people.
I didn't archieve this goal.
:-(

It's hard to explain new ideas in "monologue mode".
Do you have any suggestions how to improve readability of my site?

Blindman:
So basically, you have a series of nodes, with paths between them that have certain weights. The trick of determining a germane relation to the series of inputs the robot just received, then, will be to find the collections of nodes that relate to the input, that have the highest total weight to their paths.

I believe in school we called this the Traveling Salesperson Problem. The main problem with this is that the complexity of the problem is N!, which grows very quickly as N grows.


Intelligent system doesn't have to find the best solution.
Intelligent system has to find good solution.
Do you feel the difference?

In order to find good solution the system doesn't have to take into consideration all possible paths.
For instance, only direct relations could be considered.
It's easy to find couple of "heaviest" <a href="http://www.dennisgorelik.com/ai/Relation.htm">relations</a>, isn't it?


In fact, this "Traveling Salesperson Problem" is not a problem at all.
It's more interesting to look how AI system <a href="http://www.dennisgorelik.com/ai/Learning.htm">learns</a>.

Example:
1) AI system decides that concept #775 is the most desirable goal (just because concept #775 has the highest <a href="http://www.dennisgorelik.com/ai/DesirabilityAttribute.htm">desirability</a>
(It is simple to find concept with the highest value of desirability field, isn't it?)
2) AI system activates concept #775.
3) AI system activates manipulator #775 because concept #775 represents this manipulator.
4) Something happens outside of the AI system.
5) Sensor #437 receives a signal.
6) Concept #437 is activated.
7) AI system increases weight of <a href="http://www.dennisgorelik.com/ai/CauseEffectRelation.htm">cause-effect relation</a> "#775 -> #437".
That means AI system will consider concept #775 as a cause of concept #437.
8) Let imagine that hardcoded goal HG_3 dislikes when concept #437 is activated.
As a result, AI system will decrease desirability of concept #775, because #775 is the cause for undesirable concept #437.
As a result, probability of activation of manipulator #775 is decreased.

Dennis, last I checked this was what we called a neural network/expert system. If these were sufficient to represent human-like AI we would have had such intelligent computers years ago.

'Real AI' in VBnet is just a funny group of words that should never be used together.
As for 'real AI' in C#... well it's more likely...

What do you think is wrong with the VB.NET? And what do you mean by "C# more likely"? You probably missed the key part of the .NET concept. All the .NET languages (including VB.NET and C#) share the same cross-language compatibility and inheritability and the compiled code is using the very same CLR. The key difference between various .NET languages is mainly on the syntax level.

The key difference between various .NET languages is mainly on the syntax level.Hence the root of the problem. It fairly difficult to program complex algorithms in VB with they are needed in C.

who said it would be complex?

I think it is pretty simple actually

Dennis, last I checked this was what we called a neural network/expert system. If these were sufficient to represent human-like AI we would have had such intelligent computers years ago.

My ideas are based on expert system ideas. But my ideas are not limited by traditional expert system approach.

Could you specify, where could I read about an expert system which learns based on it's own goals? The expert system which learn accomplishes <a href="http://www.dennisgorelik.com/ai/Experiment.htm">experiment</a>?

The key difference between various .NET languages is mainly on the syntax level.
From personal experience VB.net is slower then C#. It has to do with the way the two languages are compiled into op codes.
It fairly difficult to program complex algorithms in VB
I don't see why. In fact its easier to code VB.net then C#.

exactly what i think, however i guess that is taste.

Expert Systems that learn are not expert systems any more. I came up with a new addon for my AI. Previously i merely used Empirical data to strengthen the percentage or likliness. Whilst Studying math i read about the Elementa books by some greek who's name i can't remember at the moment. What it said had nothing to do with AI but it said that Empiricial data could be relaced by a forumla instead to prove the case.
If you have 10 triangels and all have the angle sum of 180 then it is safe to assume that all triangels have the angle sum of 180 although of course an AI should be able to prove that the angle sum is 180 and if it could not then it should assume that it is "probably 180".

From personal experience VB.net is slower then C#.

That's what happens when you use the compatibility library. Stay away from it, compile the code and check it with the ILDASM. If you used the right code then you should get the exact same result.

..its easier to code VB.net then C#.

I agree with Baal on this. It's a matter of taste.

i know who you are so i know that i should agree with you but in my mind i get the feeling that VB is good to start with, however the professionals prefer C#.
When using it i get the feeling of more structure becuase of the "{", "}" but i prefer VB because to me it is more pure. It is not as messy. Think it is easier to read. But then again i started VB at the age of 7 and not C# till last year so maybe it is just something that you have to get used to.

Any works, they are compatible.

Actually we have decided that it will be hard to use .NET and we agree that our own language is in order. But we will probably make it in .NET first since we are too few, and have too little time.

I don't see why. In fact its easier to code VB.net then C#.It's easier to code... sloppily (which is usually fine). If you want to do something complex the 'simplicity' of the VB syntax just gets in your way.

anyone who is interested in recieving more information before joining can send me a PM or mail me at svenssonrikard@spray.se and i shall provide you with more information.

If you are not interested in joining then don't bother mailing me.

Physics, anyone? Would need someone with some knowledge about physics (in the field of knowing physical values (such as gravity)) (and preferebly programming skills too!)

We don't value knowing the values as much as the ability to find values on the web along with the ability to in a clever way create new values accordingly to the data you will be given.

(does it sound too fuzzy?)

Well, send me a mail and i will explain it for you.

Physics anyone? Please it is of high importance.

/Rikard Svensson

Baal: Try www.meetup.com. If lucky, you can even find a specialist who lives close to you. I had the luck a few times. When you are there, do not miss http://ai.meetup.com.

Need 2 persons.

1) Someone who can help me with physical enumerations.
2) A Neural Net VB/C# programmer. (we need to use conventional NN for our recognition algorithm)

Dennis,

I design strong AI and have a web-site ..<a href="http://www.dennisgorelik.com/ai/overview.htm">http://www.dennisgorelik.com/ai/</a>
Any questions are very welcome! :-)

Only 5 WordIDs (4 real words) per phrase? How about keeping just a single WordID field in the Phrase table and adding an extra unsigned int sequence number field to track the order so you would have N rows (/WordIDs) per phrase with the same PhraseIDs and unique sequence number? How do you want to teach the system? Did you think about the user interface? How would you tell to your AI that "Mr Black wore a black shirt yesterday” ? How does it deal with unknown words? Can your "strong AI" learn how to play games like Tic-Tac-Toe? What's the structure of the sub-system which provides the CauseNeuronID and EffectNeuronID for the Cause-Effect relation table?

Only 5 WordIDs (4 real words) per phrase? How about keeping just a single WordID field in the Phrase table and adding an extra unsigned int sequence number field to track the order so you would have N rows (/WordIDs) per phrase with the same PhraseIDs and unique sequence number?


I thought about composing <a href="http://www.dennisgorelik.com/ai/Phrase.htm">phrases</a> out of N words.
But this is wrong because of two reasons.
Reason #1:
Long phrases are not typical. That means that long phrases are used very rarely. That means that long phrases have very weak association with other concepts.
That's why it has no sense to remember long phrases.

Humans also don't remember long phrases.
Typical phrases consist of 2-3 words. 4 and 5 words phrases are relatively rare. So I think that phrase lenght should be restricted to 3 or 4 <a href="http://www.dennisgorelik.com/ai/Word.htm">words</a>.
(Not extended to N words as you proposed).

Reason #2:
From database performance point of view it's better to keep phrase in one record (PhraseId, WordId1, WordId2, WordId3) instead of keeping phrase in multiple records:
(
PhraseId, 1, WordId1
PhraseId, 2, WordId2
PhraseId, 3, WordId3
)

How do you want to teach the system?
There are two basic stages in teaching the system:
Stage #1: AI developer (I or you) programs "learning routines" and <a href="http://www.dennisgorelik.com/ai/SuperGoal.htm">super goals</a>.
Stage #2: AI System self-learns by using the learning routines. Learning process is directed by the <a href="http://www.dennisgorelik.com/ai/SuperGoal.htm">super goals</a>.

Example #1 of learning routine:
Chat with an <a href="http://www.dennisgorelik.com/ai/Operator.htm">operator</a>
After AI response to operator, operator rates AI's answer as good or bad.
AI makes conclusions what is good and what is bad.

Example #2 of learning routine:
Chat with a regular user. The longer user stays in chat - the better AI evaluates AI's actions.
AI makes conclusions what things are good to say and what things are better not to say.

Dennis,

I would not worry about the performance issues considering this particular table. It's just a bunch of numeric values (very few per row). You may have many rows but If well indexed (which you have to do anyway), the performance will be relatively OK (assuming you will use a decent DBMS). You will probably often need to select PhraseIDs of phrases which contain certain WordID. Considering the current design, that query would take 5 conditions (instead just 1) and more work for the DB engine. What I proposed would also avoid the nulls in short phrases and it would add flexibility. I'm sure you ideas will go through many changes during the development so the flexibility is generally very important. When working on AGI, keep the DB design as dynamic as you reasonably can. My understanding is that you have at least a good idea on how to convert users text-input into 4- word phrases. That's something I would like to see. What if they talk about relatively long mathematical formulas? How would you break it? Also note that having simply VarChar(20) for all words is not enough. I could never tell it that I used to code for the PriceWaterhouseCoopers (=22 chars) and not all strings in sentences are regular words. Teach it chemistry and you will find the VarChar(20) pretty short. Again, you can use a sequence and break "words" into more rows if necessary or use multiple word tables with various limits (or both ideas combined). Of course it adds some overhead but that should apply just to the very initial and the very final stage of the internal processing (+ considering the chat-like communication, relatively limited number of input/output words (at the time) would go through that process.

I would not worry about the performance issues considering this particular table. It's just a bunch of numeric values (very few per row). You may have many rows but If well indexed (which you have to do anyway), the performance will be relatively OK (assuming you will use a decent DBMS). You will probably often need to select PhraseIDs of phrases which contain certain WordID. Considering the current design, that query would take 5 conditions (instead just 1) and more work for the DB engine. What I proposed would also avoid the nulls in short phrases and it would add flexibility.

I worry about performance, because there will be more than 100000 phrases in phrase dictionary.
There will be two typical operations on <a href="http://www.dennisgorelik.com/ai/PhraseDictionaryTable.htm">phrase dictionary table</a>:
#1) Find PhraseId by sequence of WordIds.
#2) Find Phrase by PhraseId.

In "One phrase - one record" model these operations are obvious and fast.
In "One phrase - N records" model these operations are less obvious and essentially slower. Especially operation #1. How are you going to get phrase which consist from sequence of three word IDs: WordId1, WordId2, WordId3?
How are you going to build SQL query? In particular "Where" condition?

You probably keep in mind operation #3:
Find all phrases which have specified WordId.
But this operation is not typical and will be executed rarely (if executed at all).

You are right: "One phrase - one record" model waste some space for Null values. But "One phrase - N records" model waste some space for additional key (SequenceNum) values.
So, these models are ~equal from "hard disk memory consumption" point of view.
Moreover, with "3-words in a phrase" restriction "One phrase - one record" model use hard disk space much better.

When working on AGI, keep the DB design as dynamic as you reasonably can.
What is "AGI"?

My understanding is that you have at least a good idea on how to convert users text-input into 4- word phrases. That's something I would like to see. What if they talk about relatively long mathematical formulas? How would you break it?

Here is "Phrase search solution":
Imagine 10-words sentense:
Word1, Word2, ..., Word10.
For any 10-words sentence there will be
10 words
9 two-word phrases
8 three-word phrases
7 four-word phrases
6 five-word phrases
(I don't consider phrases longer than 5 words)
=====================
Total: 40 Text units in the <a href="http://www.dennisgorelik.com/ai/TextUnit.htm">TextUnit</a> collection

So when I read a text I just consider them all.

Some phrases are incidental and would not occur again in other places.
Such prases will be <a href="http://www.dennisgorelik.com/ai/Forgetting.htm">forgotten</a> (removed from phrase dictionary) gradually.

So, for me it doesn't matter if text is a regular text or it's a mathematical stuff.
However I don't think about mathematical text - it very specific and not very common form of communication.
Usefullness of mathematical notation is quite limited also :-)

If you like to know how I separate words then read:
<a href="http://www.dennisgorelik.com/ai/2004_02_01_aisdevelopment_archive.html#10764254561 6307750">Reader implementation --- Parse statement</a>
This description is out-of-date a little bit, but it will give you main idea.

Also note that having simply VarChar(20) for all words is not enough. I could never tell it that I used to code for the PriceWaterhouseCoopers (=22 chars) and not all strings in sentences are regular words. Teach it chemistry and you will find the VarChar(20) pretty short.

PriceWaterhouseCoopers is definetely not a word. It's phrase.
People treat such "words" as phrases.
Why should AI treat it as a word?

If AI text parser feels that current substring is too long then it tries to find other types of separators (including change from uppercase to lower case).

If there is no separators then word is cut at 20 chars. Not big deal :-)

Again, you can use a sequence and break "words" into more rows if necessary or use multiple word tables with various limits (or both ideas combined). Of course it adds some overhead but that should apply just to the very initial and the very final stage of the internal processing (+ considering the chat-like communication, relatively limited number of input/output words (at the time) would go through that process.

1) Yes, AI text parser should be able to find short words.

2) I don't think that "multiple word tables" is a good idea.
Multiple tables drammatically increase AI operational logic.
Human's brain doesn't work this way.

How would you tell to your AI that "Mr Black wore a black shirt yesterday” ?

"Mr Black" and "a black shirt" are quite typical <a href="http://www.dennisgorelik.com/ai/Phrase.htm">phrases</a> with completely different meaning.

Also "Black" and "black" are different.

Did I answer your question?

How does your AI deal with unknown words?
Unknown words are added into <a href="http://www.dennisgorelik.com/ai/WordDictionaryTable.htm">WordDictionary table</a> and into <a href="http://www.dennisgorelik.com/ai/ConceptTable.htm">Concept table</a>.

Also multiple records are added into <a href="http://www.dennisgorelik.com/ai/CauseEffectRelationTable.htm">Relation table</a>.
These <a href="http://www.dennisgorelik.com/ai/Relation.htm">relations</a> point to another <a href="http://www.dennisgorelik.com/ai/Concept.htm">concepts</a> which were used together with unknown words.

From this point unknown word is not unknown anymore. :-)

What's the structure of the sub-system which provides the CauseNeuronID and EffectNeuronID for the Cause-Effect relation table?

I don't understand your question.
Do you mean: "How does AI discover what is the cause and what is the effect?"?.

Can your "strong AI" learn how to play games like Tic-Tac-Toe?
Right now I don't have "strong AI", therefore it cannot play :-)

Moreover, AI prototype which I'm trying to build now wouldn't have two important features which make learning games (or other activities) possible.

These features are:
1) Ability to build it's own<a href="http://www.dennisgorelik.com/ai/SoftcodedRoutines.htm">softcoded routines</a>
2) <a href="http://www.dennisgorelik.com/ai/MotivationSystem.htm">Motivation system</a> (<a href="http://www.dennisgorelik.com/ai/SuperGoal.htm">super goals</a> and <a href="http://www.dennisgorelik.com/ai/SoftcodedGoals.htm">softcoded goals</a>).

If you like, we could discuss these features.

Dennis: sorry to take so long to reply.

Having read your description of this AI, it appears to be a combination of two things:
1) a lexical analyzer
2) a database with a series of weighted phrase-concept associations.

I'm not yet sure what the weightings are for exactly, but I see a major problem with your AI; its analysis of language has no sense of meaning whatsoever, and an extremely poor sense of context. For instance, your five-word phrases system would be able to draw some context out of the sentence:

"I had gone to the place where she had gone, but she was gone."

But none at all out of:

"I had gone after her. I went where she had gone. She was gone."

So, what does this AI do other than rate phrases by popularity?

Your AI appears to be a combination of two things:
1) a lexical analyzer
2) a database with a series of weighted phrase-concept associations.


You are right, but not exactly :-)

1) a lexical analyzer
Yes, my AI does text analysis. But this analysis is not about search for verbs, nouns, and adjectives.
Intestead my AI parser is looking for meaning of every <a href="http://www.dennisgorelik.com/ai/TextUnit.htm">text unit</a> in a text.
Find a meaning of a <a href="http://www.dennisgorelik.com/ai/Word.htm">word</a> means to find all concepts which are related to specified word.

You might want to ask: "How The AI parses is going to find this relations?".
The key to the answer is: "If words are used together, then the words probably are related to each other".
The more words occur together the stronger are <a href="http://www.dennisgorelik.com/ai/Relation.htm">relations</a> between these words.

In addition to looking for "http://www.dennisgorelik.com/ai/Concept.htm">concepts</a> relations in one sentense, my AI parser is going to search for relations between concepts in different statement.
It's implemented by using short memory.
Information about concepts which were used in previous sentences will be kept in <a href="http://www.dennisgorelik.com/ai/ShortMemory.htm">short memory</a> for some time (until it will be forgotten).

You might want to look at the core reading algorithm here:
<a href="http://www.dennisgorelik.com/ai/ReaderPrototypePlan.htm">http://www.dennisgorelik.com/ai/ReaderPrototypePlan.htm</a>


2) a database with a series of weighted phrase-concept associations.

Actually relation table is about concept-to-concept associations.
That means that relations could be:
word-to-word
word-to-phrase
phrase-to-phrase
word-to-<a href="http://www.dennisgorelik.com/ai/DeepConcept.htm">deep concept</a>
<a href="http://www.dennisgorelik.com/ai/DeepConcept.htm">deep concept</a>-to-phrase
... and so on.

In addition to that I'm going to implement in my AI reader prototype feature which relates to read/write ability only:
<a href="http://www.dennisgorelik.com/ai/TextPairs.htm">text pairs</a> mechanism.
Text pairs allow to embed "correct text intuition" into AI.

Jiri is a busy man so i will just fill in, AGI means Artificial General Intelligence. Something Jiri is extremely keen on. :)

Let me just point out one thing. It seems to me that many believe that a chat bot is the way to achieve true intelligence (real AGI)

So, you teach it a couple of words, let it ask someone a question using those words. It has encountered a problem and it askes the human operator who has the answer how to solve the problem in question. The operator tells you how to solve it and you solve it. Do you call that AI at all? Because i sure don't.

Another example. It has basic and grammatic knowledge, it begins chatting and some friendly human operator tells it lots of new data (which of course is correct). The chatbot has now got new data and it asks the operator how to solve a particullar problem. The operator gives enough hints for the problem to be solved and the chatbot solves it. You call that AI? I do not call that AI, sure as hell not Strong AI.


What i call AI is when it figures out how to solve a problem without any assistance.

My design will allow us to start our AI with a handfull of parameters (less than 10 at the moment) and it will be able to learn how to interact.
It will of course need a Goal too. The algorithm is built in a way that makes sure that it always tries to obtain as much knowledge as possible (identify everything unknown) unless a superior goal supercedes that ability.
A goal that allows it to interact is for instance "Provide your familiy" and sub-goals are that the familiy needs food, water and wood (fire)

That is what it gets to know in our "Republic".
We will also tell it how to walk. I have designed a solution so that it can learn how to walk by itself too but it would take time to perfect it so we have choosen not to.

Dennis: sorry to take so long to reply.
I see a major problem with your AI; its analysis of language has no sense of meaning whatsoever, and an extremely poor sense of context. For instance, your five-word phrases system would be able to draw some context out of the sentence:

"I had gone to the place where she had gone, but she was gone."

But none at all out of:

"I had gone after her. I went where she had gone. She was gone."


1) Well, I claim that my AI has very strong sense of context.
:-)
I explained it in my previous post: my AI keeps track of relations between concepts.

2) From my AI point of view there is no big difference between these two examples which you provided.
My AI parser would just keep information about concepts occurred in previous sentences in the <a href="http://www.dennisgorelik.com/ai/ShortMemory.htm">short memory</a>. Therefore it will be able to find relations between concepts in these several short sentences.

3) At the same time handling pronouns (like "she", "he", "it") is not an easy task for AI.
It's not easy for humans also :-)

Nevertheless my model should be able to handle pronouns.
Let's see how it works:
- After reading huge amount of text (including extensive chat) AI would know that "her" and "she" words are strongly related.
Also AI would know such facts as:
"Woman" and "she" is strongly related.
"Man" and "he" is strongly related.
"Cow" is related to "it" stronger than "cow" <a href="http://www.dennisgorelik.com/ai/Relation.htm">related</a> to "she".
"Mary" is related to "she" stronger that "Mary" related to "it".

Such knowledge would help to correctly understand sentences like:
"Mary milked a cow. She was tired."
:-)

Let me just point out one thing. It seems to me that many believe that a chat bot is the way to achieve true intelligence (real AGI)

So, you teach it a couple of words, let it ask someone a question using those words. It has encountered a problem and it askes the human operator who has the answer how to solve the problem in question. The operator tells you how to solve it and you solve it. Do you call that AI at all? Because i sure don't.


You miss the point that operator teaches AI only for limited time. All other time AI makes decisions by itself.
Of course AI has to <a href="http://www.dennisgorelik.com/ai/Learning.htm">learn</a> a lot from operator's tutoring.

Another example. It has basic and grammatic knowledge, it begins chatting and some friendly human operator tells it lots of new data (which of course is correct). The chatbot has now got new data and it asks the operator how to solve a particullar problem. The operator gives enough hints for the problem to be solved and the chatbot solves it. You call that AI? I do not call that AI, sure as hell not Strong AI.


What i call AI is when it figures out how to solve a problem without any assistance.

Ok, lets consider another example: Baal Zebul reads new information in this topic and asks his friends what does it mean. After short discussion Baal clearly realized what does this new information mean. Then Baal continued reading.
Could Baal Zebul be considered as an intelligent system (after he used help of another intelligent system)?

If I was started with no knowledge about Language, then yes.
If i was started with a basic language that allowed me to learn more then No.

I simply see it as a excuse for not being able to solve the real problem.

Why is "Mary" more related to "she", when cows are always female?
How does the parser relate seperate sentences?

haha, you are so right BigBlueHead.

That is why i say that the previous pattern is neccesary in order to understand the future pattern/layers and current.

Btw, how does it deal with this:
I have a Cow.
I have a Wife named Maria.
She is married to a Bull.

:) Which She?

If I was started with no knowledge about Language, then yes.
If i was started with a basic language that allowed me to learn more then No.

I simply see it as a excuse for not being able to solve the real problem.

It seems that you don't classify yourself as an intelligent system :-)
You definetely use help of others for learning :-)

haha, you are so right BigBlueHead.

That is why i say that the previous pattern is neccesary in order to understand the future pattern/layers and current.

Btw, how does it deal with this:
I have a Cow.
I have a Wife named Maria.
She is married to a Bull.

:) Which She?

Of course both human reader and AI reader would be confused.

Does it really matter what does "she" mean in this gibberish example?
:-)

Dennis, Dennis, Dennis. The human reader is not confused but is willing to accept several possibilities. For instance:

1) Baal has a human wife named Maria, and owns a cow that is married to a bull.
2) Baal is a bull (Baal Thebull, haha), who has a wife named Maria, and also owns a cow.
3) Baal has a wife named Maria who is also a cow and is married to a bull, whether it's him or not.

Which of these is the truth? Humans may weight the possibilities, but they also wait for more information.

How would your AI react to nonsense strings? Or, I guess the real question is, how does it connect phrases to real-world occurrences?

Ok, tell me what are the differences between "confusion" and "willingness to accept several posibilities"?
:-)

AI would also be ready to accept several possibilities.
Moreover, computer can handle multiple possibilities much easier than humans.

About handling "nonsense strings":
As I said I'm not going to put selfprogramming ability into first AI protopype.
So first prototype will simply leave multiple meanings as they are and handle all meanings together. (It's not very efficient though).

More advanced AI would be able to self-program its own respons for different kind of situations. In case of nonsense strings it would select the most appropriate program, like humans do:
- continue reading;
- reread and try to find more exact sense;
- stop reading current text at all (as you always do with my web-site when it's too hard to understand :-)).
- ...

All this choices wouldn't be programmed by the <a href="http://www.dennisgorelik.com/ai/Creator.htm">creator</a>.
Instead AI will would learn the choices with a time.

Dennis, Dennis, Dennis. The human reader is not confused but is willing to accept several possibilities. For instance:

1) Baal has a human wife named Maria, and owns a cow that is married to a bull.
2) Baal is a bull (Baal Thebull, haha), who has a wife named Maria, and also owns a cow.
3) Baal has a wife named Maria who is also a cow and is married to a bull, whether it's him or not.

Which of these is the truth? Humans may weight the possibilities, but they also wait for more information.

How would your AI react to nonsense strings? Or, I guess the real question is, how does it connect phrases to real-world occurrences?

Dennis, when I checked your website it was a series of dictionary definitions that all pointed to one another. It was a series of self-referential entries containing almost no information, which is why I originally wondered if the website itself had been created by a bot.

The point with the cow and bull example is that a human assessing the statement would seek other information to verify which of the interpretations is correct. Your AI cannot seek information from other contexts because it only parses text.

Dennis,
The best fencer in the world knows how to beat the second best because he knows what he can expect. It is the amature that might do something "Stupid" and kill him.

You will probably mess this up in your head and think that it has something to do with your AI.
But, i mean that as a Human i am intelligent because i can do a long-shot. Basically, i can use my data wisely (even though someone thinks it is stupid).

Dennis, what you are suggesting does not classify as AI. You are suggesting that your AI will become the protoge of the master and when he is ready he will beat the master.

Let me put it like this.
You have some words here and the ALF knows what they are. It puts them into a sentence (it has exactly enough data to do that) and it asks someone much smarter which gives an answer. It learns more data from that answer accordingly to the pre-programed grammar it has.
It then goes on and solves a task with that new data.

That is not AI, the only thing that is that you have taken a previously Pre-programmed algorithm and shifted the data into databases. Unless you start it with NO knowledge i cannot see how it can come off as AI at all.
But hey, Einstein said that Human stupidity is static so why not :p

Why is "Mary" more related to "she", when cows are always female?
How does the parser relate seperate sentences?

Try google query: <a href="http://www.google.com/search?sourceid=navclient&ie=UTF-8&oe=UTF-8&q=Mary+she">Mary she</a>
7.360.000 references

Try google query: <a href="http://www.google.com/search?hl=en&lr=&ie=UTF-8&q=Cow+she">Cow she</a>
1.760.000 references.

Do you see now why?
:-)

Dennis, when I checked your website it was a series of dictionary definitions that all pointed to one another. It was a series of self-referential entries containing almost no information, which is why I originally wondered if the website itself had been created by a bot.


What is your point here?
That my site is hard to read?
I know that. :-(
Do you have any ideas how to improve readability of <a href="http://www.dennisgorelik.com/ai/overview.htm">my site</a>?

Try google query: Mary she
7.360.000 references

Try google query: Cow she
1.760.000 references.

Do you see now why?
:-)


My wife is fat as a Cow. She is named Mary.

? ? ?

Is She refering to Wife or Cow? A human would say Both, but you are saying that your AI would choose Wife over Cow.

The point with the cow and bull example is that a human assessing the statement would seek other information to verify which of the interpretations is correct. Your AI cannot seek information from other contexts because it only parses text.

1) Theoretically system could be able to work only with text and be very intelligent.
For instance, our discussion with you is about parsing text on both sides (you and me) :-)

2) I'm going to extend communication abilities of my AI gradually.
In addition to text parser some other <a href="http://www.dennisgorelik.com/ai/PeripheralDevices.htm">peripheral devices</a> could be added.
These devices would be able to work with text, sound, video images, tactile input, ultrasound, temperature input and so on.

I'm going to implement text parser first of all because it has tremendous advantage over other peripheral devices.
Text parser is relatively easy to implement and it provides access to huge text knowledge base (books, Internet) which human civilization have been developed.

1> My wife Mary is terrible sick. But i have my Cow.

2> How is she by the way?

k, google might tell you that the She is Mary but honestly. The state of Mary has already been disclosed so 2 is talking about the Cow of course.
:) Better luck next time.

Sorry, Baal, I skiped some of your answer because I don't see what is your point there...


That is not AI, the only thing that is that you have taken a previously Pre-programmed algorithm and shifted the data into databases. Unless you start it with NO knowledge i cannot see how it can come off as AI at all.


You personally (as intelligent system) started with some preprogrammed mechanisms and knowledge.
Also you got HUGE amount of knowledge from humans' society.

Do you classify yourself as intelligent system or not?

For instance, our discussion with you is about parsing text on both sides (you and me) :-)

This is an intentional oversimplification and you know it. Talking is one of the last skills that humans learn, after many other experiences and skills are developed, and you're trying to put it first.

I don't know how would you undestand this text, but from this example I would understand that "Mary" is the name of your wife.

Moreover - it's not as critical if AI wouldn't be able to understand such text correctly, because humans have different opinions regarding meaning of this text :-)
Why do you expect from AI more than from humans? :-)

My wife is fat as a Cow. She is named Mary.

? ? ?

Is She refering to Wife or Cow? A human would say Both, but you are saying that your AI would choose Wife over Cow.

Dennis, i started with some basic parameters and the ability to move a little (i think).
My parameteres where food, sleep i think.

Not, "How" , "Do" , "I" , "Obtain" , "Food".

I had to figure out how to obtain Food myself. Of course i had reflexes so that when it was time to feed me i just did my pre-programmed movement for this.

That is the difference between my AI and your "so called AI".
What we humans have i think is precisly within the boundries of Intelligence. Although i see no other way to make it better. I can only do a remake.

Yes, I'm trying to implement reading skills first, because my AI will live in 21 century in my office, connected to Internet. Not in stone cave 1 million years ago.

Human beings have a lot of skills which are not required in modern world. Why should I implement all these skills in my <a href="http://www.dennisgorelik.com/ai/AIS.htm">AI</a>?

This is an intentional oversimplification and you know it. Talking is one of the last skills that humans learn, after many other experiences and skills are developed, and you're trying to put it first.

So all differences are about environment.
You were raised in physical environment, with needs to eat and to breath.
My AI would live in different environment, so its needs (<a href="http://www.dennisgorelik.com/ai/SuperGoal.htm">super goals</a>) will be different.

I'm just going to skip unnecessary skills in my AI implementation.

Dennis, i started with some basic parameters and the ability to move a little (i think).
My parameteres where food, sleep i think.

Not, "How" , "Do" , "I" , "Obtain" , "Food".

I had to figure out how to obtain Food myself. Of course i had reflexes so that when it was time to feed me i just did my pre-programmed movement for this.

That is the difference between my AI and your "so called AI".
What we humans have i think is precisly within the boundries of Intelligence. Although i see no other way to make it better. I can only do a remake.

Ok, it's time to have some fun :-)

Both I and advanced AI would be a little bit surprised about statement #1.
How would we answer will depend on our tactfulness.
So answers could vary from: "What do you mean?" to "Why do you mix your wife and your cow?" to "Does your cow worth your wife?" :-)

Less advanced AI would nicely talk about both: wife sickness and a cow.
May be it would also touch topic of "how sickness of wife affects cow's name" :-)

You cannot be sure what strong intelligent system keeps in mind (<a href="http://www.dennisgorelik.com/ai/Memory.htm">memory</a>)
:-)


1> My wife Mary is terrible sick. But i have my Cow.

2> How is she by the way?

k, google might tell you that the She is Mary but honestly. The state of Mary has already been disclosed so 2 is talking about the Cow of course.
:) Better luck next time.

Honestly, your "AI" it is just A. Artificial. It has no Intelligence.

It is not smarter than a Dictionary.

But, Dennis go ahead and make this "AI", means that there will be less opposition for me and i have never liked opponents, because i can't stand failing.

Ok, it's time to have some fun :-)

Both I and advanced AI would be a little bit surprised about statement #1.
How would we answer will depend on our tactfulness.
So answers could vary from: "What do you mean?" to "Why do you mix your wife and your cow?" to "Does your cow worth your wife?" :-)

Less advanced AI would nicely talk about both: wife sickness and a cow.
May be it would also touch topic of "how sickness of wife affects cow's name" :-)

You cannot be sure what strong intelligent system keeps in mind (memory)
:-)

First of all, your AI does not have the capabilities to say "What do you mean?"
Secondly, Your AI is lesser advanced.

First of all, your AI does not have the capabilities to say "What do you mean?"


Why not?
My AI will be able "to steal" this magic phrase "What do you mean" directly from Baal Zebul post.

My AI will apply this phrase in following conversations. It will be done absolutely mechanically, without any intelligence at all ;-)

Oh yeah, it won't be intelligent at all.
It will just follow preprogrammed <a href="http://www.dennisgorelik.com/ai/HardcodedRoutines.htm">reflexes</a> and <a href="http://www.dennisgorelik.com/ai/Goals.htm">goals</a> (like you follow yours).

;-)

Honestly, your "AI" it is just A. Artificial. It has no Intelligence.

It is not smarter than a Dictionary.

I like opponents. They help to think better. They help to find my mistakes. They point me to better options.
If you don't like opponents then it's another reason why your research will fail.

But, Dennis go ahead and make this "AI", means that there will be less opposition for me and i have never liked opponents, because i can't stand failing.

Dennis,

I'm in trouble with time. I did not read everything here and I currently cannot address everything what I would like to. I recommend you to do some performance testing with "randomly" generated data. It shouldn't take more than a few minutes to create a table with hundreds of thousands rows. I don't think it's a problem when you have roughly 500000 rows per 100000 phrases (each taking relatively few integers). We are talking about several MB only. The flexibility of the design is what's IMO important. I'm just thinking about the query you asked about. Unfortunately, I'm nearly in a sleep mode again.. You are using the SQL Server, right? My very first thought is something like:

Select PhraseID from phrase where
(seq = 1 and WordID = WordID1) or
(seq = 2 and WordID = WordID2) or
(seq = 3 and WordID = WordID3)
group by PhraseID having count(*) = 3

Would it work for you? Well, maybe one more condition to handle the count a bit better. If you do this, make the key from PhraseId and seq and add an index for the WordID field. There are also other ways how to do this.. Some work better on one DBMS, some on another.. Indexes can do lots of magic but it's definitely good to think about the needed queries before designing tables. I did not have that info before. Do you always know the positions of the words you are looking for (the 1,2,3 here)? If not then I guess you already know why I asked. If the 5 words is a limit you want to keep and if you do not want to use my seq suggestion then you may want to consider having 5 phrase tables (with a different number of WordIDs) and no nulls. But feel free to keep your current design if you like. There are other things to focus on as some of the others already pointed out.

When the system is going to forget a phrase which holds the last reference to a word, is it gonna forget the word as well? What tables are referencing words (except the Phrase)? I know you already mentioned some, but do you have the whole DB schema in a single document? If you created the DB, can you generate the DB script and send it to me (G71ai@aol.com) ?

BTW I do not think that the knowledge about "what words are usually together" will really help to get the meaning. You should probably demonstrate how the understanding and problem solving works. Would it be even able to correctly answer questions like "what's the color of a blue ball?" or "what's the speed of a car which is moving 50 miles per hour"?

IF your system is not being designed to handle mathematical tasks then I would not call it a "Strong AI" project as you did.

The "Reader implementation --- Parse statement" link does not work (at least for me). But I'm currently connecting from a very unusual environment so it could be just my local problem.

If the "PriceWaterhouseCoopers" is a phrase then how would you classify the "screwdriver"? If I understand correctly, you would break the "PriceWaterhouseCoopers" to 3 (or 4) words, right? If so, what then tells the system that it needs to put these 3 words back together when generating the output. My mind takes it as a single-word company name (but of course it can break it just as the "screwdriver"). Some languages have LOTs of these combined words. Did you think about the multi-language support?

FYI, in my AI design, I'm thinking not only about the multiple word-tables (with various limits) but also about the built-in ability to create new tables. When you just need to get an ID of a word (when dealing with multiple word tables) then it does not necessarily "dramatically increase AI operational logic". The word2ID and ID2word conversion is (as far as I understand) a very small part of the process. It will do most of the thinking using the numeric ID. And keep in mind that doing things the way how brain works (let's pretend we know how it works) is not necessarily the best way how to do it (especially when using the whole different "platform"). My favorite note on this is: "Airplanes don't flap wings". There are many things which our brains cannot handle well. That's one of the key reasons why we need powerful AI systems.

The question about CauseNeuronID and EffectNeuronID was basically "Where/how do you get these numbers?".

Do you think it can learn to recognize jokes?

What's the most impressive thing your AI currently does?

If you already addressed some of the stuff I mentioned then please ignore what I wrote. As I said, I did not have a chance to read everything. I'll be back B-]..

A.I in VB LOL

darktr00per, I call my programming language Raven. At the moment it is just Pseudo Code but i hope to create that language in the future.
I find VB very OOP and since it is compatible with C# i see no reason why not to use those 2 at the moment.

Dennis, your AI cannot quote anything from this discussion. It has no idea what it means.
And if it knows what it means then you have made it to pre-programmed. Even helping it by telling it how to phrase a question to someone with more knowledge, that is too much pre-programmed.

It should start with no knowledge at all. The parameters i was talking about were physical parameters. Etc Force = 1 m2/s
It can figure out that by itself really easy with machine vision. But it would probably be Force = 9.81 m2/s then and it would take time for the neural net to perfect that value.
We can start it without the ability to walk too.
We can start it without any goals too.
We can start our AI with a empty DB. Yes, that is right. Just the Algorithm.

Tell me when your "AI" can be started with no knowledge and i will raise my image of you as a "Strong AI" designer.

G71, thank you for the ideas about building phrase table. They are useful.
However I still think that "one phrase in one row" is optimal solution.
Even if solution which you propose is relatively fast, it is still 3-5 times slower than "one phrase in one row".

In order to read 10-words sentense you need to make 40 phrase queries.
Essential advantage of AI over humans is speed, but if we apply 3-5 times slower solution then it makes AI almost as slow as human reader :-)

Multiple tables are also less efficient because of additional space required by PhraseId key in every table and because of time lost for join.

I recommend you to do some performance testing with "randomly" generated data. It shouldn't take more than a few minutes to create a table with hundreds of thousands rows. I don't think it's a problem when you have roughly 500000 rows per 100000 phrases (each taking relatively few integers). We are talking about several MB only. The flexibility of the design is what's IMO important. I'm just thinking about the query you asked about. Unfortunately, I'm nearly in a sleep mode again.. You are using the SQL Server, right? My very first thought is something like:

Select PhraseID from phrase where
(seq = 1 and WordID = WordID1) or
(seq = 2 and WordID = WordID2) or
(seq = 3 and WordID = WordID3)
group by PhraseID having count(*) = 3

Would it work for you? Well, maybe one more condition to handle the count a bit better. If you do this, make the key from PhraseId and seq and add an index for the WordID field. There are also other ways how to do this.. Some work better on one DBMS, some on another.. Indexes can do lots of magic but it's definitely good to think about the needed queries before designing tables. I did not have that info before. Do you always know the positions of the words you are looking for (the 1,2,3 here)? If not then I guess you already know why I asked. If the 5 words is a limit you want to keep and if you do not want to use my seq suggestion then you may want to consider having 5 phrase tables (with a different number of WordIDs) and no nulls. But feel free to keep your current design if you like. There are other things to focus on as some of the others already pointed out.

When the system is going to forget a phrase which holds the last reference to a word, is it gonna forget the word as well?


<a href="http://www.dennisgorelik.com/ai/Word.htm">Word</a> usually has its own meaning. That's why I'm not going to delete word just because all phrases with this word are deleted already.
However word could also be <a href="http://www.dennisgorelik.com/ai/Forgetting.htm">forgotten</a>.

And if word is forgotten then all phrases with this word have to be removed from the phrase dictionary table.
In other hand - why would AI want to forget word with is actively used in existing <a href="http://www.dennisgorelik.com/ai/Phrase.htm">phrases</a>?

What tables are referencing words (except the Phrase)?

<a href="http://www.dennisgorelik.com/ai/WordDictionaryTable.htm">Word dictionary table</a> is referenced by <a href="http://www.dennisgorelik.com/ai/PhraseDictionaryTable.htm">phrase dictionary table</a> and by <a href="http://www.dennisgorelik.com/ai/TextPairs.htm">Text Pairs</a> table.

Also <a href="http://www.dennisgorelik.com/ai/WordDictionaryTable.htm">Word dictionary table</a> refer to <a href="http://www.dennisgorelik.com/ai/ConceptTable.htm">concept table</a>.

BTW I do not think that the knowledge about "what words are usually together" will really help to get the meaning.

Making conclusions based on "what words are usually together" would help. It wouldn't be enough though :-)
But it would help a lot, because this "close words" analysis is fast and can use huge text base (from Internet, for instance).

A.I in VB LOL
Are you programmer?
Are you familiar with Turing machine?

he probably meant LOL as Lots of Love instead of Laughing out Loud :p

Dennis,

Just a note on the seq query in case you will consider using something like that in the future development:
On current machines, it should not take more than something between 0 and 25 milliseconds per a "seq" query (running against half a million records) if the phrase exists (and possibly a bit more if it does not exist). The way how you build those "40" queries can let SQL Server significantly benefit from cached (sub)queries so some of the subsequent queries might take almost no time + with certain algorithms you may be able to skip some queries considering the results from some of the previous queries.

Questions:
Doesn't your system have a problem with yes/no questions?

What exactly the system does when it "activates concept"?

You mentioned learning from the Internet.
What if the text_to_learn_from contains something like
"X is broken. [a few more sentences] X is blue.. X cannot do Y."
Is the system gonna understand that the X (if not broken) might be able to do the Y? Note the ambiguous "blue" and the possibility of making an incorrect relationship between the "blue" and the "cannot do Y".


"Find a meaning of a word means to find all concepts which are related to specified word."

..Which (for your system) means to find all the words which relatively often appeared close to that word, right? How about concepts like "he" or "it". You cannot copy the Internet into your DB. If you keep just the "top" 100000 phrases then it might be just lots of junk data. You need to get the meaning and generalize most of the knowledge. It's not very clear to me how you can generalize with the level of "understanding" which your system is capable of.


"If words are used together, then the words probably are related to each other."
True, but not necessarily in the sense of cause/effect. Also note that there are many causes and effects which cannot be described in a few words. It often gets a lot more complex.

Your google example (on the word counts) may not be very valuable because it does not say how many times are the words used on those pages and if they are close enough to each other to allow your system to pick the relationship up.

I recommend you to do some experiments with reasoning (including things like deduction, induction, abduction, analogy analyses etc.)

BTW in order to think well, your system also needs to have some sense about things like time, space, minds of other subjects etc.. And put some thoughts on system's creativity..

Doesn't your system have a problem with yes/no questions?


I don't know the answer yet :-)
My system is still in "design and development".

But the design of the system would allow it to self-learn with a time. The learning will include "how to understand specific questions" and "how to respond to specific situations.

I'm not going to hardcode "Yes"/"No" constructions.

It seems that you do hardcode it. And here you come into trap of getting immediate results right now and sarcrifizing universal cognitive abilities in long term.

I even described this approach here :
<a href="http://www.dennisgorelik.com/ai/FalseWaysInAISDevelopment.htm">False ways in AIS development</a>


But you know - by your approach you even doesn't get immediate results:
checking if AI could solve these "yes/no questions" is not more than a game for the users.

It doesn't have much of business sense.

Typical users' questions would be: "Tell me about that" or "How could I achieve this?".

What exactly the system does when it "activates concept"?


When the system <a href="http://www.dennisgorelik.com/ai/Activate.htm">activates</a> a <a href="http://www.dennisgorelik.com/ai/Concept.htm">concept</a> it just simply the concept into fast memory (RAM). I name this memory: <a href="http://www.dennisgorelik.com/ai/ShortMemory.htm">short memory</a>

Short memory keeps current thought: set of concepts. For instance, 100 concepts.

These <a href="http://www.dennisgorelik.com/ai/ActiveConcept.htm">active concepts</a> could be handled in a special way.
For instance:
The longer these concepts are in the short memory together, the stronger their inter-<a href="http://www.dennisgorelik.com/ai/Relation.htm">relations</a> become.

Another specific is that active concepts could cause AIS <a href="http://www.dennisgorelik.com/ai/Action.htm">action</a> with higher probability than non-activated concepts.

You mentioned learning from the Internet.
What if the text_to_learn_from contains something like
"X is broken. [a few more sentences] X is blue.. X cannot do Y."
Is the system gonna understand that the X (if not broken) might be able to do the Y? Note the ambiguous "blue" and the possibility of making an incorrect relationship between the "blue" and the "cannot do Y".

1) The father the sentences are the weaker <a href="http://www.dennisgorelik.com/ai/Relation.htm">relations</a> between these sentences would be.
2) Even if the system remembered wrong relations from one example, later on this problem would be fixed.
Just because majority of examples will will teach correct relations between concepts (words, phrases).
3) So my thinking is that this relatively "linear" <a href="http://www.dennisgorelik.com/ai/ReaderPrototypePlan.htm">reading</a> will provide good results.
4) BUT!
In order to achieve good understanding it is not enough just to read the text.
The system has to act. Make several choices about whether to continue reading or to read again or apply different technique to reading the text or just drop the whole reading.
5) That means that <a href="http://www.dennisgorelik.com/ai/DecisionMaking.htm">decision making system</a> and <a href="http://www.dennisgorelik.com/ai/MotivationSystem.htm">motivation system</a> have to be implemented.
I'm not going to fully implement these systems in the first AI version.

Dennis Gorelik: "Find a meaning of a word means to find all concepts which are related to specified word."
..Which (for your system) means to find all the words which relatively often appeared close to that word, right? How about concepts like "he" or "it". You cannot copy the Internet into your DB. If you keep just the "top" 100000 phrases then it might be just lots of junk data. You need to get the meaning and generalize most of the knowledge. It's not very clear to me how you can generalize with the level of "understanding" which your system is capable of.

1) You are right, but not completely :-)
The system doesn't have to find ALL related words.
Just the strongest relations have to be used.

2) Instead of copying the Internet into my database I will just copy information about words and most common relations between the words.
Unusual phrases will be regularly removed from my <a href="http://www.dennisgorelik.com/ai/MainMemory.htm">database</a>.
See: <a href="http://www.dennisgorelik.com/ai/Forgetting.htm">Forgetting process</a>.

Dennis Gorelik: "If words are used together, then the words probably are related to each other."
True, but not necessarily in the sense of cause/effect. Also note that there are many causes and effects which cannot be described in a few words. It often gets a lot more complex.

Your google example (on the word counts) may not be very valuable because it does not say how many times are the words used on those pages and if they are close enough to each other to allow your system to pick the relationship up.

I recommend you to do some experiments with reasoning (including things like deduction, induction, abduction, analogy analyses etc.)


This relatively simple mechanism ("if words are together --- they are related") moves system understanding in right direction.
Whole intelligence is based on this "use every clue for moving in right direction" mechanism.
"If words are together --- they are related" mechanism is not good enough for strong thinking process.
But this mechanism is vital for generating knowledge background for an <a href="http://www.dennisgorelik.com/ai/IntelligentSystem.htm">intelligent system</a>.

In addition to that, easy to medium intelligent tasks complexity could be handled by using this mechanism.

That's why I'm going to implement "If words are together --- they are related" mechanism in the first version of my AI.

BTW in order to think well, your system also needs to have some sense about things like time, space, minds of other subjects etc.. And put some thoughts on system's creativity..

I agree with you only about TIME thing. Feeling of time has to be embedded into the system.
Space sensors can be added to the system later.
Understanding of other minds will come with experience (like reading, chat, ...).
System creativity will emerge based on <a href="http://www.dennisgorelik.com/ai/MotivationSystem.htm">motivation</a> and experience (<a href="http://www.dennisgorelik.com/ai/Read.htm">reading</a> and <a href="http://www.dennisgorelik.com/ai/Experiment.htm">experimenting</a>)


------------
But once again: I'm not going to implement all this stuff in the first version.
It's very important to select only critical features for development. Otherwise the development will last forever :-(

Dennis,

What would your system do when asked "What's the current euro/dollar exchange rate?" ?

Can you show a User-AI sample dialog (including some questions and answers) ? I know it's not working yet but can you just make it up based on your expectations?

Dennis,

Here is an example of a business oriented y/n question: "Does the functionality F represent the only difference between systems S1 and S2?"
I’m getting lots of y/n questions when I do my IT and RE business. It’s definitely not “game for users” specific.

You made some comments about my AI and the syntax. Unfortunately I currently do not have time for creating a readable document about my User-AI communication module to give you a good sense of how it works. Let me just tell you that it's not just a language with certain syntax. It's more like a communication environment which includes tools which let users to build very dynamic named concepts which then can be used in sentences. From system's point of view, particular word is never just a string, there is always a defined concept behind it, so the system really knows what it means. The syntax of the sentences (which are using dynamic concepts) is currently mostly hard coded but it's not necessarily a problem (I'm currently analyzing its expression power). In certain sense, human languages are not really changing from the practical point of view. It has something to do with the relatively static set of senses we have. I did put some thoughts on the syntax editor and the possibility of making it available for some categories of users but I (maybe just temporarily) left the idea, mainly because it was causing some nontrivial challenges for algorithms which are involved in sharing knowledge between independent copies of the system and because it would require users to know more than what I expect them to know.
I'm using the syntax because I do not know about any other reliable way how to get meaning from sentences with a decent accuracy. Your "blind" approach to the meaning IMHO cannot work and getting the meaning using known NLP techniques does not work well either (even though it's much better than what you are doing). The ability to get meaning through NLP has a tremendous importance for governments of all advanced nations but because of it's complexity, it stays not_well_solved after 50 years of active research. Certainly, I'm not planning to spend my lifetime working on that. The actual AI algorithms are a lot more important to me. The Novamente group claims some level of indispensable success in NLP but so far, they did not really demonstrate any NLP related breakthrough. I got the specification of their internal language (the "KNOW") which I believe was developed based on CycL. That helped me a bit when thinking about my communication module but I have done most of the stuff differently.

Dennis,

I recommend you to put some thoughts on more powerful AI designs. Simple question-answer systems are just too limited (even when working well). BTW there are already some q/a systems (with NLP) which work relatively OK (even able to answer the euro/dollar question correctly). Note that the NLP is not that difficult when working just with relatively simple questions. But that’s not enough. We need AI which can think about complex scenarios. You would say for example: Imagine a scene where ..[a long description of a complex hypothetical scene, possibly including some non-standard rules which do apply there].. And then you ask for steps how to solve a particular problem in that theoretical scenario. The solution might take many (well scheduled) steps and the system should be able to generate the steps when it's theoretically possible considering the available information.

Here is an example of a business oriented y/n question: "Does the functionality F represent the only difference between systems S1 and S2?"
I’m getting lots of y/n questions when I do my IT and RE business. It’s definitely not “game for users” specific.


I agree, this is not game question.
But what will happen if you ask 8-years old student this business question?
Will he/she really answer it?

Do we agree that 8-years old student is a good example of a <a href="http://www.dennisgorelik.com/ai/StrongAI.htm">strong AI</a>?

My point here is that it's practically impossible to build super-advanced intelligence without building moderate complexity intelligent system in the process.

After moderate complexity AI will be build we will be able to correct our models, remove some AI features and upgrade other AI features.

I'm using the syntax because I do not know about any other reliable way how to get meaning from sentences with a decent accuracy.

1) Natural language is a good approved tool for "meaning exchange". Why do you need to invent something else? Key to understanding of natural language is finding relations between words/phases/other structures in the sentence with knowledge base of the <a href="http://www.dennisgorelik.com/ai/IntelligentSystem.htm">intelligent system</a>.
2) Why do you need "decent accuracy"? Humans are able to solve complex problems without decent accuracy in their mental process.

Your "blind" approach to the meaning IMHO cannot work
What do you mean here under "my" approach?
Why do you name it "Blind"?

I agree, this is not game question.
But what will happen if you ask 8-years old student this business question? Will he/she really answer it?
If the F, S1 and S2 concepts are relatively simple then yes. My 3 years old daughter can do that.


My point here is that it's practically impossible to build super-advanced intelligence without building moderate complexity intelligent system in the process.

You are extracting information incorrectly in the initial data processing. If you use advanced AI algorithms (developed later) against that KB then you still get the "gigo" effect.


Natural language is a good approved tool for "meaning exchange". Why do you need to invent something else?
Because (considering the current technology) learning and understanding NL is not the same for humans and machines.


Why do you need "decent accuracy"? Humans are able to solve complex problems without decent accuracy in their mental process.
I was talking just about "getting meaning from sentences". No machine can currently do it as well as an average human. I would call it "decent" when machines can understand what we can.

What do you mean here under "my" approach?

The design of your "AI" system.

Why do you name it "Blind"?

Because it cannot see and understand the stuff which is important for reasoning.

DennisGorelik: "What do you mean here under "my" approach?"

The design of your "AI" system.

Because it cannot see and understand the stuff which is important for reasoning.

Do you consider <a href="http://www.dennisgorelik.com/ai/Experiment.htm">experiment</a> as a part of my AI design?

Please review <a href="http://www.dennisgorelik.com/ai/EventsCorrelationAnalyzer.htm">Event correlation analyzer</a> article also.


Do you still think that my AI design with these features won't be able to reason?

You are extracting information incorrectly in the initial data processing. If you use advanced AI algorithms (developed later) against that KB then you still get the "gigo" effect.


Why do you think that advanced AI features will use knowledge base created by simple <a href="http://www.dennisgorelik.com/ai/TextParser.htm">text parser</a> only?

Quite contrary: advanced AI features will read the texts again (or even read new text). Also other sources of information might be used.

For example, when you as a developer are trying to solve a new problem - do you always restrict your research by knowledge which you already keep in your head?

Do you consider <a href="http://www.dennisgorelik.com/ai/Experiment.htm">experiment</a> as a part of my AI design?


No


Please review <a href="http://www.dennisgorelik.com/ai/EventsCorrelationAnalyzer.htm">Event correlation analyzer</a> article also.
Do you still think that my AI design with these features won't be able to reason?

Yes

Dennis Gorelik: "Do you consider experiment as a part of my AI design?"
No
Dennis Gorelik: "Do you still think that my AI design with these features won't be able to reason? "
Yes


Ok, why it won't be able to reason?
<a href="http://www.dennisgorelik.com/ai/Experiment.htm">Experiment</a> is the base for humans' reasoning. Why wouldn't it work for AI?

Why do you think that advanced AI features will use knowledge base created by simple <a href="http://www.dennisgorelik.com/ai/TextParser.htm">text parser</a> only? Quite contrary: advanced AI features will read the texts again (or even read new text). Also other sources of information might be used.

I'm sorry, I just did not expect your system reading everything over and over again when it gets a bit more clever. I recommend you to forget the "simple" text parser and focus on more advanced parsers. I see too many problems with your current algorithms. You talked about learning from Internet. Many pages do not contain text where you could find causes and effects (might be just lists of features of various products, etc..). When it does contain some causes and effects then the order is not always "cause..effect". It can be as easily effect and then cause. My understanding is that that would be pretty confusing for your system. The causes and effects will be often represented by much longer texts than what you showed as an example on your page and a distance of these parts of text can just be too long for your system to handle. And how about time related issues? What was true yesterday (or years/centuries/.. ago) might not be true today. In fact it could be very incorrect. The years old and currently invalid info might be on many more pages than the latest info which should be used.. Especially the technical, political, finacial, medical, etc.. info. Many pages are also not really describing our world (it's about fiction etc..) + what's true for one subject could be false for another regardless time/location. The relativity plays a significant role in understanding this world. Your simple idea will not work well and 100K phrases would not be enough anyway considering the way how you generate it. The order of parsed articles may easily have a very negative impact on the "forget concept" functionality which would potentially give you a very different result comparing to the same scenario with a many times bigger phrase table. There are many more problems. I intentionally mentioned just some of those which might be closer to your current mindset.

For example, when you as a developer are trying to solve a new problem - do you always restrict your research by knowledge which you already keep in your head?

No, but I do not have to experience everything again. I might just reorganize some of the knowledge (if necessary considering the new experience).

I'm sorry, I just did not expect your system reading everything over and over again when it gets a bit more clever.

Why not?
Humans do this all the time.
As far as I know, iterative approach in <a href="http://www.dennisgorelik.com/ai/Learning.htm">learning</a> works the best.
Why shouldn't AI use the same iterative approach?

Simple <a href="http://www.dennisgorelik.com/ai/Concept.htm">concepts</a> and <a href="http://www.dennisgorelik.com/ai/Relation.htm">relations</a> are studied first. Then learning of more complex concepts is based on knowledge of simple concepts and relations...

I recommend you to forget the "simple" text parser and focus on more advanced parsers. I see too many problems with your current algorithms. You talked about learning from Internet. Many pages do not contain text where you could find causes and effects (might be just lists of features of various products, etc..).


List of features and products is a great source of <a href="http://www.dennisgorelik.com/ai/CauseEffectRelation.htm">cause-effect relations</a>.

If products are located at the same page then they are most probably related to each other. They are <a href="http://www.dennisgorelik.com/ai/Cause.htm">cause</a> and <a href="http://www.dennisgorelik.com/ai/Effect.htm">effect</a> for each other.

Let's imagine that: "monitor" and "processor" are posted at the same page. Then AI will infer that "monitor" is a cause for "processor" and "processor is a cause for "monitor".
Later on such cause-effect relation could be useful when the AI will try to solve a problem "whether this shop which sells processors can sell monitor also".

I agree with you that simple text parser will not highlight complicated relations between concepts. Complicated relations will be lost among simple relations. Even if complicated relations are far more important. But simple text parser will provide huge knowledge base with basic common sense. That's the purpose of the <a href="http://www.dennisgorelik.com/ai/ReaderPrototypePlan.htm">simple text parser</a>.

More advanced (and essentially stronger!) cause-effect relations will be discovered in <a href="http://www.dennisgorelik.com/ai/Experiment.htm">experiment</a> and by more sofisticated routines (including advanced readers, chatters, knowledge downloaders designed for specific subject, ...)

No, but I do not have to experience everything again. I might just reorganize some of the knowledge (if necessary considering the new experience).

You might avoid re-reading, but quite often you re-read old sources in order to solve new problem.
And you do so, because re-reading old sources is the most efficient way to solve the problem.

Why shouldn't it be the most efficient way for AI?

When it does contain some causes and effects then the order is not always "cause..effect". It can be as easily effect and then cause.

That's why sequence: Word001, Word002 will increase
Word001 -> Word002 relation by 2.0
and
Word002 -> Word001 relation by 1.0.

Constants 1.0 and 2.0 are selected for demonstration purpose only. <a href="http://www.dennisgorelik.com/ai/ReaderPrototypePlan.htm">Text parser</a> might work the best with other values for these constants.

The causes and effects will be often represented by much longer texts than what you showed as an example on your page and a distance of these parts of text can just be too long for your system to handle.
This task will be accomplished by advanced readers.
BTW, majority of humans have problems with inferring from text parts located too far from each other. :-)

And how about time related issues? What was true yesterday (or years/centuries/.. ago) might not be true today. In fact it could be very incorrect.

Common sense doesn't have to tell what is correct.
Common sense have to tell what could be correct.
Common sense points to possible solutions.
Also common sense identifies "suspicious/completely unknown" ideas.
All these things are very helpful in reasoning.

<a href="http://www.dennisgorelik.com/ai/ReaderPrototypePlan.htm">Simple text parser</a> will generate exactly this common sense.

The relativity plays a significant role in understanding this world. Your simple idea will not work well and 100K phrases would not be enough anyway considering the way how you generate it. The order of parsed articles may easily have a very negative impact on the "forget concept" functionality which would potentially give you a very different result comparing to the same scenario with a many times bigger phrase table.

I'm lost here...
I don't understand why my simple idea wouldn't work with 20K <a href="http://www.dennisgorelik.com/ai/Word.htm">words</a>, 100K <a href="http://www.dennisgorelik.com/ai/Phrase.htm">phrases</a>, and 10 million <a href="http://www.dennisgorelik.com/ai/Relation.htm">relations</a>?

I also don't understand about "negative impact" on the "<a href="http://www.dennisgorelik.com/ai/Forgetting.htm">forget</a> concept" functionality...
Could you clarify your points here?

Dennis, you got some hints here and you got some by email. I think we could keep explaining and arguing for a while but is it really worth it? We might be just wasting our time considering how easy it is to develop a simple test version which would tell you more than anyone here. Your basic ideas, however correct or incorrect from the AI perspective, seem to be clear enough to be coded - that's one of the few things I like about that ;-). Many other AI thinkers are just dreamers without a good sense of the current technology. Ask yourself what did you do since you started to discuss it here (or since you designed the part of the DB we talked about). Your simple prototype should be IMO already going through the testing stage and you should be already able to see for yourself what some of the real problems are. I believe that a useful test version can be written in several days by a single hot shot developer. Make it simple and let it fill the DB with some data. You have mentioned that you will use the multiple-phrase-table design as I recommended some time ago (which is good for the real version) but you might want to use your original idea to keep this test version as simple as possible (you can convert it to more tables later). The test version (of course) does not have to learn from the Internet etc. Let it learn from text files in certain folder. Use all shortcuts you can think of. Keep focus on getting the data into your simple DB. The algorithms to support dialogues or problem solving on the user-interface level do not have to be supported in this version. The related queries/algorithms can be (I guess) tested manually (relatively few tests could then show you how it works). If you still want to spend some time with the theoretical stuff then just make sure you understand definitions of various types of reasoning and put some thoughts on these (http://www.goertzel.org/papers/AGIQuestions.htm) questions.

If you still want to spend some time with the theoretical stuff then just make sure you understand definitions of various types of reasoning and put some thoughts on these (http://www.goertzel.org/papers/AGIQuestions.htm) questions.

What is your version of answers to these questions?

Did you read how do other researchers answer to this questions?
(I couldn't even find answers of the author of these questions...)

Dennis,

What is your version of answers to these questions?


That document is not done yet. It takes time to explain things well. So far I had other priorities than explaining it to others + I did not make final decisions about some algorithms. But thinking about the questions can help to design interesting AI systems.


Did you read how do other researchers answer to this questions?


Yes, some did find time for explanations. Most of them were very young and relatively inexperienced AI designers who like to keep things secret. I do not think it was really necessary in those cases but I do respect the conditions we agreed on.


(I couldn't even find answers of the author of these questions...)

They did answer it.. I'm not sure where the document currently is and if the answers still reflect the current design of their system but I did read it about a year ago. It's somewhere.

BTW FYI, I recently got a ridiculous amount of long AI related emails and I'm pretty behind with answers. But I do read everything and I'll answer some of your questions when I get to it.