I am new to A.I and i am a visual basic programmer and have very in depth knowledge about this language. Please tell me can i use this language to create intelligent softwares.If no then which language i have to learn for creating intelligent systems.

As i am a beginner please tell me about the following things

1. Programming Languages for A.I
2. A.I Internet resources

AI ain't language dependent. Actually, it has nothing to do with programming or programming languages. AI is all about algorithms and concepts that you can play with on paper. So, their "digitalisation" in computer programming is independent of the language.

As for Internet resources... browse this site or use Google.

Dear Eric,
Thanks a lot to reply my question. But please tell me, you mean that by using A.I concepts i have to create alghorthims on paper then i can apply them in visual basic.

Sort of...

You know, AI is a _WIDE_ subject. What are you trying to do exactly?

Dear Eric,
Thanks Eric you have provided a great help to me. you ask me what i want to do. Simple but very complicated to apply i want to create an intelligent system which automatically learn from the internet about new things, technologies, ideas and everything. As you know as we grew old we learn new thins from our daliy life. I want to make this system like this. TO create this type of system in which A.I fields i have to study and which concepts i need to know.
Many people say that A.I concepts are easy to implement in an Object oriented language like C,C++ , Small Talk, and VB.Net . Are they saying right.

For creating this system whih i explain above from where i have to start and more i tell you that i hate neural networks.

That sounds like an extremely difficult thing to attempt as an early project.

Neural networks, as they currently exist now, probably wouldn't be much good. I would suggest you look into the area of knowledge representation. Looking at how languages such as LISP work would probably help; even if you do not use a language like this the ideas involved would be helpful.

To make a system that can REALLY learn about new things from the internet you would probably have to make some sort of breakthrough in A.I. You may be able to make a system that fakes it, though.

In which A.I fields i have to study to create an intelligent system which automatically learn new things from internet. And tell me is it important for me to learn logic programming with Prolog.

hello, i am new member. :) I am a IT student. now is my final years, and i need to create a intelligent system. so i get the trouble. which programming language is suitable for create a intelligent system, and how to create ?? can anyone teach me ?

There are many programming languages developed for AI, but I don't use any of them. I use Xbasic which is broadly similar to, but (I beleive)more powerful than VB. You can download a copy of this open source software and manuals if you join Yahoo group: Xbasic.

On Yahoo group Xbasic I have just posted a Personal Dictionary Creator. This is a foundation AI program. It uses some passive learning concepts, which is what you seem to be interested in.

The PDC works by scanning all your files for character coded a to z and A to Z. These form about 20% of all 256 possible bytes.

Rules:

1 list all filespecs (files and paths) on every drive.

2 load 1Kb (1024 bytes) of data from each file (or the whole file if smaller)

3 analyse the bytes - Regardless of the file xtension: .txt .doc .htm etc. , if a file has more than 20% of alphabet characters then it most probably contains readable textual information.

4 if more than 20% text, save the filespec to a new list.

5 read every byte of every file in the new list (by loading strings)

6 if seven or more chunks of text are separated by whitespace and terminated by a full-stop, treat this as a sentence - save each new occurence of a word to an array - count occurrences.

7 sort the list in order of frequency of occurence.

8 the passive learning bit -

you can use the word list to discover the favourite topics of the computer user - words in that topic will be high in the list and will tend to form clusters.

The accuracy of the PDC is about 99% for most systems, i.e. it only collects about 1% of garbage.

As might be expected, the word "the" is usually the top scoring word on English language based systems.

By the way, try the ai forum at www.ai-forum.org - they have 4 main forums and you can try out a "Hal" conversational program.

Try a hal - play rough - find the defects - write a snippet of program which would avoid the problem.

As for what to study - I recommend at least an introduction to linguistics - most especially computational linguistics.

Don't forget: for any topic you can usually search for an FAQ page.
This is often better than a dictionary when it comes to strange terminology.

Good luck!
Patrick

To Ehsen .

The first thing to do is to make a spider for the web.How to do that?
you make a programm witch
1. you enter a start URL address.
2. Your programm downloads the page.
3. It searches it for trigger words eg "calculus" , "AI" , etc
4. It searches for links and it puts them in a stack , to download them in the next step.
5. it Keeps a log for the resuts (pages) or the whole document.
6. it keeps a IP map of its paths.
You search for standard HTML tags for links in each page.Do you know HTML ?

alexander from Hellas (GR)

A list of tasks on how to do something is not AI and especially not
real intelligence. The brain i.e. mind is a problem solver. It is given a goal, some blocking from the goal, and then it goes its memories and algorhythms to find solution. Telling how to solve the problem is not AI. It must have the capacity to think. Integrate the known, mock up a course of action, and then within reasonable certainty decide whether the probability of an acceptable outcome is greater than the risk.

None of the current AI languages are worth anything. None of them are truly intelligent, and the whole in not sentient.

If you want to truly program AI, forget everything you have been told and try something else.

jb.

Hi jb

> None of the current AI languages are worth anything. None
> of them are truly intelligent, and the whole in not sentient.

AI languages are not intended to be intelligent. They just make it easier to develop AI applications.

AI languages like Prolog, Lisp and Smalltalk make it easy to develop code quickly - i.e. they are good prototyping languages.

You can write AI applications in C++, Fortran or Pascal; it's just takes longer.

GL7

We will be introducing CCI or Crytal Cyber Intelligence. This will not be intended to be artificial but the real thing. This is a great forum and we are getting A LOT of suggestions, many which are helpful and will be implemented. CAIO will think, control nano and robotic applications and is pretty ambitious.

We now have funding and a lab to work in and a growing number of team members. I will post the CAIO page again when we get things together. So much to do.

jb

I have a great idea that I recently started working on..
The AI I am creating interacts with a user by becomming familiar with the words the user enters.
It starts with no knowledge of language, so any language can be taught to it.

You enter text, as if you are talking to a baby, then it will ask you simple questions about the words such as "noun?" or "adjective?"
Of chorse, these are present in every language, and the words are substituted by pictures. For example, "adjective" is reprisented by an image showing different colors, an arrow, and a texture. The noun is reprisented by the picture of a person and earth. Get it?
As you teach it about the world, it will begin to pick up on how to use the words and even how to read them. For example, when you've taught it much, it can identify and answer your questions or even ask or solve its own complicated questions. It also can "learn" math.

The program is much more complex than this, of chorse, but it remembers each word you teach it and several things that define that word as "noun", "adverb", "pronoun", "gesture", "verb", etc.

In response to a topic above, I thinkit is impossible to create an AI that can learn by its self from the internet, because there is no one to teach it the truth or identify the nouns and verbs for it. Also, there's alot of non-sense on the internet and many different languages. Besides, do you really like the idea of a program thinking it's smarter than you?

Back to my post, do you think this will be a successful project?
Written in Microsoft Visual Basic...

luopa

I started with this idea back in 1991 or thereabouts. It is good training, and your basic idea is sound, but!

Infants don't learn by being taught about nouns, verbs etc. They learn passively, that is, without any formal training in the use of words.

I think they do this by latching on to cues in every utterance. Cues are mainly what might be called grammatical words, but every word known to the hearer/reader is a cue of sorts.

Example:

The rocket ship landed on Vraxz.

Every fairly literate reader will assume that "Vraxz" is a planet. There is no need to ask an "authority", or look it up in an index, you just "know". That is passive learning.

The simplest cue based parsing algorithm searches utterances using these types of templates:

a ** eos
an ** eos
the ** eos
on ** eos
with ** eos

where ** is any word and eos means "end of sentence".

In these case, even allowing for errors, there is a high probability that the word is a noun.

You can develop your own templates for each class and sub-class of noun, adjective, verb, adverb. (Primary verbs: be do have, auxiliaries and prepositions etc are "hard-wired" as cues.)

In passing, to everyone here:
I can only access the web for a few days every fortnight, so I tend to fire off a cluster of postings when I can. But I will do my best to follow up these discussions.

Cheers,

Patrick

avoid plague like the cliches!