Quote:Original post by fguihen hey mr anonymous poster. please dont call people idiots . it does not apply and the fact that , even though the problem in hand may seem impossible at present, it wont get solved if nobody talks about it. how bout we all forget about artificial conciousness and maby in a coupple hundred years it will pop up all by itself. have you ever solved anything without exploring it fully? im not joining the discussion, its over my head, but im really enjoying this post and everyones input.keep working everybody, keep reaching. if you dont try you cant succeed ( corny but true), and dont listen to this anonymous poster. he/she's the real "idiot"
If you want to learn stuff about AI, go look up research papers and learn from them. They are the ones qualified to talk about these kinds of things. You are listening to uneducated people. Just out of curiosity, how many people that have posted here have taken any courses on AI? How many intense AI projects have you worked on? Be very carefull who you get you info from, you could easilly be mislead. Not everything you read on the internet (especially forums) is intelligent.
Ap -> Pipe down. This is a discussion of How we would create a conscious entity. Its got nothing to do with how impossible it is.
Maybe we bring this down a notch. To build a resonably intellegent entity, which can possibly be conscience with further effort.
From, Nice coder
Click here to patch the mozilla IDN exploit, or click Here then type in Network.enableidn and set its value to false. Restart the browser for the patches to work.
Well, I can't speak for the others, but I've done my research, wrote papers when it was proper and designed a few complex systems - though as you can tell by some of my posts, the AI wasn't of this structure. At any rate, that reminds me - you may wish to start out with the basics of what already exists. Read up on backprop, reinforcement learning (just the simple stuff - Q-learning), genetic algorithms. The foundations of AI. A lot of times you're looking for solutions to these things, the best way to do it is by piecing together existing work - especially because most of the things that have been said in this thread so far are repetitive. Start at ai-junkie.com (especially now that ai-depot.com appears to not be quite what it used to be) to get some feel, then read papers from citeseer. There's a lot to learn out there.
Havn't read all the posts so sorry if I'm repeating a point:
Maybe we're trying to walk before we can crawl - I think that for such a project to succeed, rather than trying to create an AI that is as intelligent as a human, maybe we should try to make one that's as intelligent as some of the simpler mammals. Maybe a dog or a cat or even one of the simpler primates. I'm one of those people who beleive that animals also have a form of consciouness - if you've ever had pets you'll know what I mean. Dogs have their own personallity, their own likes/dislikes, their own way of doing things - and dogs certiantly arn't the most intelligent animals around. Sure they'll never understand quantum mechanics, or that the earth rotates around the sun, it's simply beyond their scope - but then again I'm sure plenty of people wouldn't also be able to.
I think if we were to create an entity that was able to behave in a similar fashion to a dog or a cat then we would really be getting somewhere. In fact I think in alot of areas we have far surpassed that level of intelligence. I remember watching a TV show about the cleverest pets in the UK (or something), one of the tests performed on the pets was to watch a treat fall from a set height onto a low level table...now most of the animals (including dogs and cats) expected the treat to land underneath the table - and they were pretty confused when they couldn't find it (the treat was obviously on top of the table). Surely an AI would easily be able to predict that the treat would land on top of the table?
Turns out the most intelligent pet was a budgie that had learnt to sew with a needle and thread (I'm not kidding!).
You start of simple, by adding extra objects, extra data mining algorithms, ectra links. (or letting it find those links by themselves), it just gets smarter. But you can Restrict the environment so that its optimal.
Its all about a simple system, which can be smart enough to suit the needs of many tasks. Which would have differing degrees of conciousness.
From, Nice coder
Click here to patch the mozilla IDN exploit, or click Here then type in Network.enableidn and set its value to false. Restart the browser for the patches to work.
I think something that everyone has been missing so far is creativity. I think language is a creative talent where we draw on a vast vocabulary of words with different meanings and methods for using each word to describe our streams of thought. I think the two hurdles that is challenging everyone is: 1. scientifically defining the creative process people use 2. scientifically building the mechanics of language linguistics and defining how to define a sentence
I think anyone trying to build a chat bot will have to have a thorough understanding of linguistics and how words are used together (should be language independent).
I think a chat bot or intelligent entity will need to have an initially loaded method for asking questions (predefined nodes which mutate over time, such as asking "what is XYZ?" in different languages)
I think once we can concretely define the rules we use to do something creative, we can model those rules algorithmically and reproduce the same results. I'm sure theres lots of variables, such as taste & style.
IIRC there was a neural network, that was composing music "Creatively". it was ages ago, tho.
Do you think creativity is a requirement of concienceness?
From, Nice coder
Click here to patch the mozilla IDN exploit, or click Here then type in Network.enableidn and set its value to false. Restart the browser for the patches to work.
i figured out how to do it simply. you need to do it all in parts. then interact and integrate the parts into each other. for example, one part is its vision, the other part its object regocnition database... just many many parts and many different types of links.
Is creativity required? Well, probably - it's sort of part of curiosity, which is extremely important in ever becoming truly intelligent (as you've realized). That neural music creator, though, wasn't creative, per say. They're called autopredictive - after training many times by reading over music by human composers (usually only one), they can predict what that composer would choose as the next note after a certain sequence. By feeding that prediction back in as an input, you can ask it what it would predict for the (next) next note, etc. Not really the same thing as creativity. Check out Kurzweil's neural artist for a slightly more creative program (kurzweilai.net)
Quote:Original post by jtrask Is creativity required? Well, probably - it's sort of part of curiosity, which is extremely important in ever becoming truly intelligent (as you've realized). That neural music creator, though, wasn't creative, per say. They're called autopredictive - after training many times by reading over music by human composers (usually only one), they can predict what that composer would choose as the next note after a certain sequence. By feeding that prediction back in as an input, you can ask it what it would predict for the (next) next note, etc. Not really the same thing as creativity. Check out Kurzweil's neural artist for a slightly more creative program (kurzweilai.net)
for curiousity you could just encourage it to "like" positive results. have it put more power into researching positive results. and give it an initial programmed "curiousity" that would branch out based on positive results.
