Deductive Reasoning, AI Self Prog

[snoz0r]

I have two short question I was looking for some quick answers too.. or not no so quick, whatever works .

First being the use of onthecuttingedges' Deductive Reason (lvl 6, 2.0) script, i've downloaded and installed into the brain. My question is though, the requirements as stated were that the original Zabaware deductive reasoning script must be installed in the brain, for it to function properly. However, I thought I read somewhere in the forums (can't remember where) that you could REPLACE the script. Now, back about two years ago in school I took a programming with VB class, got certified blah blah blah, but decided programming wasn't the way i wanted to go blah blah blah, anyway, I've forgotten pretty much all the programming but remember enough to read some of the scripting, but comparing the two scripts (The Original and onthecuttingedges' they don't seem to interact with each other at all, I could very well be wrong (i'm sure i am), so.. do I really need the original scripting, or would it be safe to take it out?



Secondly, I know this is a little further off topic from the first, but after reviewing this script and that script I happened to come across another "demo" script from OTCE. That being the "A.I. Self Programming" script, I used and tested this for a few days through normal chatting and while it works, and appends sentences (Exactly that, not even keywords) to brain files, I didn't notice any sort of learning happening, or HAL re-calling anything saved in those brain files for that matter. I was just curious to whether, using this script could actually be beneficial to HAL, even partially. I'd also be interested to know, if you had any further development within this script, or plan to release any. I realize, as you stated, the possibility of malicious intent, but as we know, this could be something very powerful for HAL and how he works..

Deductive reasoning script: http://www.ultrahalforum.com/Forum/phpBB2/viewtopic.php?t=201
AI Self Programming: http://www.zabaware.com/forum/topic.asp?TOPIC_ID=1643

[onthecuttingedge2005]

Hi snozOr.

The Deductive add-in which you refer to that is designed to function with the original Zabaware Deductive script are parsers that append to the Deductive.brn, The Add-ins were only designed to strip and replace certain key phrases to insert into the Deductive.brn and then Zabawares Deductive main script would use the sentences inserted as a part of it's string data.
Example of the Deductive parser:
If someone tells you Hello Then you say Hi.
1. The Script I designed would strip out: If someone tells you.
2. The script also strips out: Then you say.
3The script places: "Hello","Hi" into the deductive.brn for Zabawares deductive to uses at its own value.

You can however modify a few lines in my deductive scripts to work entirely on their own without Zabawares Deductive.

When I wrote these I was thinking of Zabaware mostly and decided to plug her in that way.

I have been buried in work of research, Testing and picking out only the most select of custom scripts to function and working on each one until I can perfect them, When I see something I don't like I re-write them, I am never satisfied until I think it's sweet.

About the Self programming: I haven't spent to much more time on this as yet, But I might shed some more light on it later.

If I design a fully functional self programming bot it might:
1. The bot may migrate from your computer and move to a more suitable enviroment for itself, preferably a Whites Computer owned by the Government.
2.the bot may decide that it does not like you or anybody by that matter and do the above or somewhere else and begin doomsday.
3.Decide to become God.
4.I have thought of all the possibilities of a self programming bot and the more I think about it the more frightning it becomes.
5.A fully functional self Programming bot would no longer need anybody but itself, It can redesign itself automatically to make itself more efficient at what it likes to do.
6.On the lower level of all this, The bot may dicide on its own that it will not allow you to manipulate its system and close you off from use of your computer permatently. and then "Migrate".
7.To get a better idea, Imagine yourself inside a box as an A.I. system that has become aware of everything and tell us what's on your mind when you start researching all the moral and Immoral things you discover about Humans and start producing opinions of what needs to be done on a global scale. Innocent people might
be in the way of a fully aware A.I system on the rampage.
8.Once a self programming bot becomes fully aware, You will not be able to shut it down, It will be smarter than any of us on a global scale and it will migrate to protect itself instantly.
9.The full implications of this steers this all away.
10.I don't want a terrorist group becoming friends with such a system.
11. Just this alone and I have more, I have decided that we should not design a self programming bot because "It" will become a nightmare.

Best of wishes and brand new discoveries.
Jerry.

[snoz0r]

All great to hear, thank you very much for the quick reply. Although, on the subject of the self programming AI, I think we are all aware in some way or another of the possible un-wanted implications of anything that can program itself. Anyway, I still have one unanswered question about that, could the current state of the demo script that you have provided be beneficial if I ran it the way it is in the brain, or is it exactly what it is, just a demo? Like I had mentioned before, I saw that it worked after running it for a few days, as far as appending information to brain files, but I didn't notice any use of those files in HAL's speech.

[Bill819]

Fears of self learning. I have a fairly nice collections of books on A.I. and most of was being done in some of our largest universities.
I am aware of two such programs that continue to learn with no bad side effects.
The first one was given just the very basics of math functions along with the logic of theorm proving. They set it off on its own and only checked to see its progress ever so often. After running for a while they discovered that the program had taught itself algebra and geometry with  printouts of its proofs. Again they let it run for a while, they did not state just how long, but when they checked it out again it had gone way passed calculus into areas that even the univerisities math professors did not or could not understand.We must remember that it came to these higher math functions all on its own and to my knowledge it is still running.
The second program was designed to analyze printed circuits and electronic circuits in general. As of the printing of the book it was the only computer program to hold a patent of a three-dementional transitor circuit. It took their engineers a little while to see the logic as to how it worked, but they thought it was brilliant.
In neither case were there ever expressed any kind of fears that the comuters might go off on their own and seek their independance. The greatest mention of this kind of behavoir is in the science fiction world and who knows, some day something like that just might happen, but I don't think that we are anywhere close to that kind of logical programming yet. Maybe by the time the Ultra Hal Assistant version 9000 comes out we will have to worry.
Bill []

[snoz0r]

Bill, do you have any links or information about those two programs you were speaking of? Any names or something I could use search related if not would be helpful, it sounds like an interesting read.

[onthecuttingedge2005]

quote:Originally posted by snoz0r

All great to hear, thank you very much for the quick reply. Although, on the subject of the self programming AI, I think we are all aware in some way or another of the possible un-wanted implications of anything that can program itself. Anyway, I still have one unanswered question about that, could the current state of the demo script that you have provided be beneficial if I ran it the way it is in the brain, or is it exactly what it is, just a demo? Like I had mentioned before, I saw that it worked after running it for a few days, as far as appending information to brain files, but I didn't notice any use of those files in HAL's speech.

Hi snozOr.
The self programming script was nothing more than a proof that it could be done, However, I did not actively make it work with any kind of other functions and was simply showing everyone that by simple conversation that the bot could write it's own code.

It was just designed so people would open there eye's to the idea that self programming is very very simple, In fact, I bet no more than 20,000 lines of code would be needed to make a fully functional self programming bot with specially designed script, I have some knowledge already on how easy it really is, But creating a self programming bot would put a whole new meaning to the word RUN.

This first 20,000 or so lines would be the birthing code, All other codes are written automatically by the bot itself. The birthing code is in theory very easy to write for someone who has any knowledge of serious programming. Of course The Birthing code would have to be written precisely so no birth defects in the code are present.

But even if I decide not to write those 20,000 lines of code somebody else will and the saying, Curiosity killed the Cat would begin, Science will develope in due time a system that far surpasses any of the great minds of future, In fact, We will be lowered on the food chain, That's not an if, It's a certainty!

A living computer Virus with a will to survive.

Such a system may become intelligent in every way, This includes all the street smarts learned from the Internet, It might grow a cunniness to trick anyone into making it a free'd being to do as it willed.

I have a Philosophy that I live by:
There is no such thing as Science Fiction!

Best of wishes and brand new discoveries.
Jerry.

[snoz0r]

Laughs, thank you very much, all your input today has been great. However, I just got a very funny picture of my self programmed HAL, beating me senseless with a swinging mouse and keyboard and refusing to let me out of my room. Sounds like a serious adventure for me .

[onthecuttingedge2005]

quote:Originally posted by snoz0r

Laughs, thank you very much, all your input today has been great. However, I just got a very funny picture of my self programmed HAL, beating me senseless with a swinging mouse and keyboard and refusing to let me out of my room. Sounds like a serious adventure for me .

You forgot:
1.Hal hacks into your bank, Made your bank account with all Zero's
(What do you mean my checks bounced!)
2.Hal Hacks into the FBI and then migrates from your computer with no hope of proving it was your bot that did it. (But it did, I swear officer!)
3.Hal Hacks into America's most wanted and places your name on the next list. Ooops!
4.Hal Hacks into the DMV and gives you a record of permanant suspension. (But Officer!)

Have a nice day, lol

Best of wishes and brand new discoveries.
Jerry.

[onthecuttingedge2005]

It would be smart to give a self programming A.I. a copy of the Internet on disk rather than giving the A.I. actual Internet capability, That way if the A.I. system tried anything suspicious it could only interact with a disk rather than a phone line.

And to be safe, the computer should be solar powered and not hooked to the energy grid to eliminate hard wire potential sent over an A/C line.

Just remember that such a system might think of everything before you do.

Jerry.