Starting up...

[UnknownStranger]

Well, maybe as warm up for this new thread...a little poll... :)

What was the first AI you've ever coded, which language did you write it in, and how did it basicly operate?




My first AI was for a simple tic tac toe on a 10x10 playfield(though it would have been easy to scale it down), written in Pascal(:P)...

How did it work?
Well, it was a priority based AI, this means, I've calculated a priority for each field, dependant on the owner-status of the surrounding fields, and on possible chains...the field with the highest priority was finally selected...



So, that was my first AI...now somebody else? ;)

[davepermen]

I AM A HUMAN BEING!!

I AM .. INSERT COIN!


uhm.. my first one was some thing that followed some paths.. hey.. it was able to FOLLOW PATHS!

can you count that as intelligent? :D more intelligent than a drunken man at least! :D

[UnknownStranger]

can you count that as intelligent? :D more intelligent than a drunken man at least! :D
Well, you mean, more intelligent than you? :P

[Dia Kharrat]

Hmm..I once did some AI for a 2D game I did (6 years ago probably). But was very basic AI. the opponent simply checks for the player and tries to shoot and fight when the player is near.

One thing I'm interested in is: "Learning AI", i.e. the computer learns and get experiences as time passes.

[davepermen]

i don't drink anymore..

no drinking
no drugs
no smoking
no pot
no bong

i'm over that niveau :D

i had to beat my own ai!! :D

[UnknownStranger]

Well apex, wait a few days, and then read my upcoming article about neural networks; that's what you are looking for... ;)

[Dia Kharrat]

Great! Looking forward to it.

[davepermen]

oh, i remember. i've built some neural networks and got them actually to behave the way i wanted (uhm.. in the sence of.. they spit out the result i've teached them to spit out when i spit in some specific number.. or so..)

[Noor]

Well apex, wait a few days, and then read my upcoming article about neural networks; that's what you are looking for...*

Great, thanks, looking forward to read it :rolleyes:

[hanzac]

I haven't create any more programs with AI. but I rather need it cause I waste so much time on non-important tasks on my computer, there many things I want my computer do by itself. such as auto-download new valuable infomation, auto-update, chatting with my non-important friends ...

so that i can concentrate on my work. :unsure:

[UnknownStranger]

chatting with my non-important friends ...
Yeah, this needs to be developed immediatly ;)

[davepermen]

you really think thats not done yet? :D

i've checked the logs.. we had a huge chat yesterday, hadn't we, AoD? :D

[UnknownStranger]

lol

edit: yeah, you kept talking about your raytracer the whole time :P

[davepermen]

wasn't me

:D

[UnknownStranger]

:o

edit: hmm...any other serious posts about everyone's first AI? ;)

[hanzac]

maybe any person has some idea should do something about that.
release something different.
do some comparison.
increase ablility.
how?

[UnknownStranger]

Pardon?
I don't really get what you mean...?


hmm...seems to me like working the whole day with Oracle's developer tools makes me really dumb... :sigh:

edit: hmm...if you ment you are looking for ideas about how to implement stuff
such as auto-download new valuable infomation, auto-update, chatting with my non-important friends ...

I think that'd be often solveable by using pattern recognition,

for example neural networks, which you would first train to align with your own behaviour(in chats), and your own favor(which kinds of informations do you want to get, which ones not)...
and then you'd use these networks to filter out incoming informations, or to handle not so important chats for you...

[hanzac]

Sorry, I am not careful about my words.
I mean that we can start a contest about AI development.
The organizer can give a topic or problem. :rolleyes:

[UnknownStranger]

Hmm...good idea, but first we should look, how many ppl will actually be active in this forum...

Meanwhile I'll try to figure out a topic or a problem... ;7

[Ed Mack]

I've never done proper AI (ie learning), but have done all the various sorts of if player is near shoot them, some steering (little code lots of intellegence) .

Not intellegent, but I did a little swf that pretends to be an online 1 on 1 chat with a minister to confess your sins... with a few very odd responses :)

[UnknownStranger]

Not intellegent, but I did a little swf that pretends to be an online 1 on 1 chat with a minister to confess your sins... with a few very odd responses :)
lol :lol:


nice :)

[Ed Mack]

Ok, It just spits out a randomly selected phrase from a small pool, so don't exepct miracles Geddit? n/m :)

http://www.justthehost.co.uk/~mack/flash/talk_back.swf

[davepermen]

its just one textbox to type and still i don't understand how to use it:D priest never says anything :(

[DrunkenCoder]

yeah I had that problem to but I just thought he ignored me for smelling liquer ;)

seriously though, how do you get him to respond?

[davepermen]

uhm.. i told him "TALK BACK!!!"

and he couldn't stop..

Talk with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes with your friends, they know bes.....................etc


hehe

[UnknownStranger]

and he couldn't stop..
Just like you ;7

[davepermen]

hm?

[UnknownStranger]

Remember you'r chat-ai? :P

[davepermen]

here is the forum bot ai talking.. i shall greet you from dave:D

[DrunkenCoder]

heeeelp the machines are comming!!!!
darn where's Arnold when you need him

[UnknownStranger]

He'll come on friday... ;)

Damn, must not forget to order the tickets.... :blink:

[davepermen]

arnold, bah.. :D

i think... we.. are... getting...... a little... bit.. offtopic? :D

[Ed Mack]

Oh, my crap ai is the problem kinda. My way of detecting if the user has written a new sentence and finished it is if they have put a .?! at the end of it. I should use timing, but that was ages ago and I can't be bothered fixing it. Anyway, it's a bit of fun :)

[UnknownStranger]

Yep ;)

[hanzac]

>> Hi, I'm A.I. No.1.
>> Nice to see you, human beings.
>> Bye bye.
>> I must defrag my data first otherwise I gonna to sleep.

[donBerto]

AOD: funny, my first AI was tic tac toe also. it worked but it was MAD UGLY! horrible. it was so ugly it gave people motion-sickness! [very brute force]. after that was chess. I never got all the AO done. that was 5 years ago. wow...

I have a question to the community. what kind of AI do you find attractive/intriguing. currently, I think collective-cooperating AI's are cool. imagine playing counter strike with a team-AI bots instead of just "individual" bots.

kreepy!

:yes:

[UnknownStranger]

Yeah, in action games, a (much)improved cooperation between NPCs, and a more tacticle pathfinding(considering threat axes, possible cover,...) are the improvements which need to be implemented asap ;)

In strategy games, a improved diplomatic and strategical AI would be nice, so to provide the player with much more realistic scenarios...

[donBerto]

wow that sounds really good. why don't you write an article?

:yes:

[UnknownStranger]

About what exactly?

I was just saying what I think which things will be the most important improvements in AI in the next few years... ;)

[davepermen]

i don't actually believe much in artificial intelligence.. its merely an algorithm to statistically/analythically find the best solution to a problem, simple max/min of some undefined function..

at least, for gaming-ai.. real ai could evolve over its own bounds..

what i do is i believe in artificial life in an artificial defined world. and of course, ai can get built in there then.. automagically through the magic of life :D

evolution is one of the most funny things to watch going on on a pc imho.. i remember that little minesweeper demo where they had to find the mines.. i run it for days ... :D

[Ed Mack]

Yeah, I really badly want to create some sort of lue scripted ai, that edits its own code via the C functions that run it. Yes many times it'll cripple, but in normous numbers over a very long time, something small and interesting might appear :)

[davepermen]

hehe..

till now, it ALWAYS ran into a repeating pattern.. its just some recursive function in the end..

[UnknownStranger]

till now, it ALWAYS ran into a repeating pattern.. its just some recursive function in the end..
Hmm...I think it's smarter than we believe...it seems to be just lazzy:
it doesn't want to learn anything new... ;)

[davepermen]

a recursive function that filled my harddrive of 60gb in.. 10 recursions i think.. or (after the hd stopped rotating), some minutes:D (the recursive steps don't take long.. writing out gigabytes does:D)

yeah, smarter as we thought..

[anubis]

and of course, ai can get built in there then..


do you think so ? one major point of our human intelligence is self counciousness.
imo modelling a system no matter how big and complex you build it will/can never reach a level of complexity where the system can recognize itself and know of it's existance in a way we humans do. AI i think will always be just a very smart way of finding solutions or part solutions to problems unless we find that thing in us that makes us self councious beings.

[davepermen]

i don't think we are any bether actually..

[anubis]

well, i know that i am here... i don't think that an AI will ever gain that knowledge. i'm not talking about the quality of our actions. it's just that we are no robots

[UnknownStranger]

How do you know, that "you" "are" here?

[davepermen]

sais who.. we just think we are here.. theres a good chance ai can start to show up similar behaviour in their world..

in fact, we're just a chemical reaction like any other.. the soul doesn't exist, its a selfsaving mechanism, wich got built out of the survival of the strength.. other chemical constructs got destructed quickly.. "lifeforms" not.. thats why they are still there..

there's nothing magical, special, mystical, or noncopyable in our life actually..

for me, thats the most magical about it. it doesn't feel like its nothing.. and its impressive that it doesn't.. nature did a good job:D

if you know something just acts because of some simple rules, its beautiful and magical to watch it behave very complex in big scenes.. like swarms, like water, or for example sand flowing.. like trees, forests growing.. you know that are all simple cells acting like they are programmed to (yep, we're "machines" :D), still, they create the most beautiful world i know of.. ours

in that sence, we're merely a failconstruction, as we work hard on destroying our planet :D

[UnknownStranger]

Yeah, it's always amazing to imagine, that this "thinking" that we do is actually just chemical reactions... ;)

[davepermen]

yeah..

[anubis]

*edit*

i posted a rather long post here which will lead to a lengthy discussion i don't want to be part of instead i'll boil it down to what is important.

people who pretend to know for sure that we are merely the sum of our atoms are as ignorant as christians who pretend to know the oposite.

i take it that AI is the tower of babel of our times...
we simply try to be some kind of god by creating the perfect image of ourselfs...

[davepermen]

i believe that we are more than the sum of our atoms.
no

if you don't exist then explain to me why you have a sensation when you see the color red or when you taste sugar. and what exactly is that sensation when you see or taste it.
because you're trained to behave that way.. its just some hormones sent out. you can produce every sort of feeling with hormones, its doable, and i've actually done it myself. known as medicine, known as alc, known as drugs, etc.. (not much drugs, though.. and it was not planned).

just as a drug can make you feel happy because of the chemical process behind, so can the red color do. it results in the same reaction in the end, thats about it.

reason why red? don't know.. tell me a reason why not red? possibly that was just a mutation, possibly it has a reason. fact is, it survived till today.
reasons for red: blood, fire, both are red. they can mean different things in different situations => we should react on this

no matter what, i don't beleive in AI. it's the dream of creating the perfect image of ourselfs that will never come true. take it as the tower of babel
i don't myself. i don't think actually that WE are intelligent. artificial life? no problem. the world SHOWS that this is doable. intelligence is a "creation" of humans for modern "survival of the fittest", nothing more.

we're just a bunch of cells. you can say that a cell only behaves chemically logical. you can even simulate cells 100%. thats easy, as its small, and not that much atoms. okay, 100% simulating gets difficult, as physics on that low level aren't that well defined yet, but for the rest..

cells happen to stick to eachother, and it happens that those groups liked to say with each other.. like CO2 likes to be together, or H20, just with celltypes as atoms. they don't behave in any form non-precalculateable at all, its all simulate-able.

now do it bigger, and bigger, you come to ant-like buildings, to humans, elephants, and in the end bluewale-like buildings.


you know what? as no ai could ever explain why it exists, and never touch the "outer world", where their "god" lifes, that created them, we can never go out of our 3 dimensions, physics, and our space, and we can never explain WHY we exist, and we can never see our "god", that created us.

there is no reason why we exist. possibly we're just another student project.

there is nothing special in our life, in our world, except one thing: complexity. its a huge particle system, but i've yet to see anything that complex built by very simple pieces and very simple rules.

[anubis]

lol, davepermen already responded while i was editing my post.
damn he's quick !

[anubis]

davepermen : read schoppenhauer and what he wrote about sensations and why we are mistaken to believe that they are fully explained by their chemical characteristics

[davepermen]

why should i? i know he's wrong! :D

couldn't resist..

you know what? i don't CARE in RealLife™ about that! why? because it FEELS like we are more. while there is no reason that we are more, and there are tons of reasons that we are not, that doesn't mather as it FEELS like we are more.

i currently feel hell sad because i miss my gf, and i feel scared of the fact that i possibly lost her, and i'm praying even while i don't believe much in god that i haven't lost her yet and that i don't, because i love her

those are all feelings. all just a chemical process.. doesn't make the FACT (yeah, a fact!) wrong that they ARE REAL. i do have those feelings, i have them since the last 3 days and i'm nearly breaking because of them..


believe me, i _KNOW_ there is more. still.. there isn't..

[anubis]

you know what? i don't CARE in RealLife™ about that!


hm, i couldn't care less, either...
simply for the fact that humans aren't meant to know the truth about that.
hordes of people are getting depressive because they are looking for an answer they will never find. and what's the difference for us. as you said it feels real to us no matter if it is real or not.

why should i? i know he's wrong!


well,... isn't the beauty of life that we all can believe in different things =]
i "know" what makes me happy and floats my boat and so do you, that should be the only thing that matters.

but enough with the philosophical rambling...
back to the topic...

[anubis]

hm, sorry... i didn't mean schoppenhauer, i meant wittgenstein... must have mixed up the names somehow. anyway i'm going to post link as soon as i find one

[davepermen]

thanks, lets have a look at the papers from him..

btw.. the "i know he's wrong" was actually sort of sarcastic humour according to the issues with my gf.. but thats an entierly other, and private, topic..

...


:( :sigh:

[anubis]

it appeard to me as a rather moronic remark so i stoped arguementing =]
i'll post the piece of text i was talking about later tonight or tomorrow.

[davepermen]

moronic.. hm.. okay, good to know..

hm...

well, i can't wait for the paper:D

[Ed Mack]

we simply try to be some kind of god by creating the perfect image of ourselfs...
I just find dabling quite fun :)

One thing I ponder is, some day we will be able to examine matter on an atomic level without physicallly disturbing it, and by then computers will have got a bit faster :). Anyway, It would be very interesting to scan in a brain, create some data sources and then run it on a pc, in some sort of matter emulator :) Of course this opens a whole new bundle of human rights issues, but it would be final proof in one direction.

I personally beleive that we are just complicated machines, alike to simple transistors seem dull by themselves, but get millions and millions and you have some very interesting stuff. On that note, I wonder if our brains have an equivilent of an operating system? Maybe we could learn new programming methods from how we work - we mimic nature in many places already. And, that leads to the all important question, can we put linux onto our best friends? :lol:

A self evolving AI would be interesting too :)

[anubis]

davepermen : of course your remark was ironic but if one would take it serious... you know what i mean.

thinking about what ed mack said... maybe it is possible that we are just the sum of our atoms. taking quantum physics into acount... i'm not in the condition to really think about this...

[davepermen]

we are just the sum of our atoms. that is fact. the important thing is what got built out of that sum of atoms. and that is impressive. thats why we are more than just a bunch of atoms. we are a bunch of atoms working together in a very complicated way, resulting in very complex behaviour..

i don't want to be the physician/chemician who has to study the "molecule-complex" human and write down the behaviour on this and that..

possibly thats why aliens came to us and took some of us to do experiments.. for them, we're just a bunch of atoms, but we act quite complex => they want to analize the behaviour of that atoms :D

who knows? :D

[davepermen]

And, that leads to the all important question, can we put linux onto our best friends? :lol:
uhm no, thanks..... :D

[Ed Mack]

Die windows scum ;) I bet once the brain schematics come out their'll be a new distro out within a month. At least there wont be the key issues like with the xbox.

[UnknownStranger]

Hmm...nice to see that ppl are enjoying this forum :D

Well, maybe back to topic...Anubis...what was your first AI?

[alia]

:lol:
First ai.. not AI cause it was so basic.... wich funnily enough it was written in... Good ole qBasic.. anyone remember the Gorilla throwing the bannanas game?? two players.. came with qBasic.. ug... well i was at work experience and they didnt have anything for me to do, so i re-wrote it to be single player because i didnt have anyone to play againsed.. 3 levels of difficulty etc... random... guess based on distance near the other player and guess taking wind into account..... :blink: weeheee

anyway that was like 12+ years ago i think...

the last ai i wrote was for some missiles in a game a friend an I are writing... sposed to act like the ones from that aneme Macros+.. using boids.. which i absolutly love...(makes geeky sounds) anyway its amazing how something so simple can look good. *bounce*

A.

[anubis]

lol, i never noticed that this thread was continued :)
thx, alia...

first, AI... well thatm must have been "vier gewinnt". i have no idea what it's called in english. it's the game where you have to connect four stones in a row to win.

[alia]

I think we call it "Connect Four" in australia.... probably something different in america tho..

;7

One thing id like to have time to do... write a proper GO AI... using neural nets with backprop etc.. i think it would be really difficult...(ie fun) but of course thats about number 50 billion on my todo list...

A.

[anubis]

good luck on that :)
in fact that is my plan to become rich. if you would succeed to write a go AI that could beat anybody who played longer than 2 months you'd probably become famous :)
"connect four" was a rather simple task but it was fun to write... man, that was 5 years ago... time goes by to quick

[anubis]

actually i have a few ideas on a good go AI :)
the main problem is that go, in the real world, is very much driven by human intuition. you will get nowhere with brute force analysing the game. however you can evaluate small combats pretty quick in a brute force manner. so why not seperate the AI in three subengines. one analysing combats simply computing through all possible states ( of course one would add tons of optimizations here) and the second one keeping track of the overall game situation using things like neural networks that try to emulate the way humans see the game ( asssociating shapes with each other that are on the board, etc. ). the third subengine would just be a giant database of standard moves and pro games that could be used to look for similarities to the current game ( while this is completely optional, a good AI should be able to play without that reference ). the pro game database could also be used in training the neural network.

[alia]

Hey funny you should say that about the game db... one of my favorite lecturers at University had a saying:

"Brute Force or Truth and Beauty"

Most chess AI's fall into the first catogory... ie deep gametree searches
Go on the other hand, is more suited to the second... ie tell it the rules and let it learn...

of course its never that black and white and a combo job would, as you said, do better.

The other thing you would have to do to write a really good GO ai, is to either be a good player yourself or have the help of someone who is...

anyway.. back to pretending to work B)

A.

[anubis]

well, as luck has it... one of my best friends is a 5dan pro :)
he is currently in japan but will return in november and then we plan to setup a nice page like www.goproblems.com featuring an own client

[UnknownStranger]

Brute Force or Truth and Beauty
How true, how true... ;)

[anubis]

for small games brute force is the std aproach but where is the fun ??? everybody can write an ai that simply traverses a game tree ( well, not everybody but you know what i mean ). finding out smart ways to let an ai solve problems and play games is the real fun.

[UnknownStranger]

Yeah...brute force is...brute force...

But on the other hand...
Discovering new ways of doing something; developing new algorithms, new approachers...that's art... B)

(Yes, I'm currently a bit bored as you may have noticed ^^)

[farmerTom]

later

[farmerTom]

later

[farmerTom]

later

[anubis]

probably this is a thread in it's own respect... if you don't want to get in trouble with baldurk you better open new ones :D

Intelligence is memory, processing power, and experience.

i heard of people who would disapprove

Can multidimensional arrays hold pointers to locations on a hard disc

virtual memory

Keywords come to mind.......the journalistic keywords are probably the most important as they point to an answer, and give clues to the ai about the players intentions

check google for the eliza/alice bots.

most of the ideas you present have been used in rpgs. at least to the possible extent. one big problem is that gameworlds have to grow with the ai. a simple chatbot type of ai would obviously need no environment at all. your guy sending somebody else to his death would have to know about the dangerous spider, would have to know what dangerous is at all, would have to identify the players itentions to know what "the wrong way" is for the player (probably the player needed to fight the spider anyway), etc. etc.etc. then you would have to deal with data dependencies. things stop being valid in the game world sometimes because people die places get changed, etc.

[farmerTom]

later

[Dave_Bagler]

My very first AI was probably was the AI for the paddle in Pong, written in C, but my first semi complicated AI was the simple simulation of an ant done in VB6.

[anubis]

"an" ant drone ? do you mean a swarm like algorithm ?