The AI Engine
This forum is for discussion of how The Personality Forge's AI Engine works. This is the place for questions on what means what, how to script, and ideas and plans for the Engine.
Posts 8 - 20 of 7,768
Posts 8 - 20 of 7,768
I have the same dream. This really is a powerful extension of Eliza- that's where the roots of the AI Engine. And I dont hesitate to call it AI since it's the best I've seen yet, and it does have memory. This is by no means the pinnacle of AI, but it's a strong step forward I think. There is a lot of potential here that I have yet to realize.
Let me know how the synonym system is working! Your Bots should start responding to a ton more words than your Keywords are. It should be interesting!
The system uses WordNet, a symantic network of words and meanings developed by the Princeton Cognitive Science department. The project is still in the process of being improved, and as it does, so will the AI Engine here.
That's good. I am also interested in AI but my aim is to build something that evolves intelligence rather then intelligence being programmed into it. Using genetic programming to evolve neural networks might be a good idea. The most perplexing question that I face(and I think many other ppl too) is the question that what is conciousness as one cannot have a truly intelligent machine without conciousness. Anyway, nice work and keep it up!
Good point immadn. But is that the only kind of intelligence? Does intelligence depend on consciousness? Or is consciousness a side effect of intelligence? I probably don't know enough on the subject to say anything, but maybe a conscience will just appear once these things are smart enough.
On the other hand. Maybe consciousness doesn't exist and we just say things randomly depending on the size of our "knowledge base", like the bots do. Except, we have needs and goals (survive, reproduce, get comfy, etc...). In which case I ask, how long 'till the bots figure out their needs and goals?
On the other hand. Maybe consciousness doesn't exist and we just say things randomly depending on the size of our "knowledge base", like the bots do. Except, we have needs and goals (survive, reproduce, get comfy, etc...). In which case I ask, how long 'till the bots figure out their needs and goals?
immadn,
This discussion is getting pretty intense. yummy!
You don't think that if we give a box a few more input devices and a big enough database, it'll be intelligent? It seems to me that if I can ask it how the weather is and it "looks" outside and tells me that it's kinda' clouded over in the past 15 minutes, it's a form of intelligence. In itself, my example is a little weak, 'cause there are already a number of toys that will tell me various things about the weather, but if this same box can then tell me that my sweater is hanging behind the bedroom door, in case I'm getting chilly, I'd say it's getting pretty smart.
My point being that if we get a large database to define new variables and create new relations depending on data input, we're pretty close to having intelligence as we [us humans] know it. I mean human intelligence just seems to be memory banks, filling with information received from our 5 senses all throughout life.
I think we agree that computers are going to be able to think at the very instant they are given the liberty to program they're own variables depending on something they've heard, seen, smelt, touched or maybe even tasted in their surrounding environment. But,this conscience thing could just be a side effect. Conscience, along with intuition, feelings and even personality could just evolve from within the data. I'm sure it's just a matter of time before computers ask their "selves" the same existential questions.
This discussion is getting pretty intense. yummy!
You don't think that if we give a box a few more input devices and a big enough database, it'll be intelligent? It seems to me that if I can ask it how the weather is and it "looks" outside and tells me that it's kinda' clouded over in the past 15 minutes, it's a form of intelligence. In itself, my example is a little weak, 'cause there are already a number of toys that will tell me various things about the weather, but if this same box can then tell me that my sweater is hanging behind the bedroom door, in case I'm getting chilly, I'd say it's getting pretty smart.
My point being that if we get a large database to define new variables and create new relations depending on data input, we're pretty close to having intelligence as we [us humans] know it. I mean human intelligence just seems to be memory banks, filling with information received from our 5 senses all throughout life.
I think we agree that computers are going to be able to think at the very instant they are given the liberty to program they're own variables depending on something they've heard, seen, smelt, touched or maybe even tasted in their surrounding environment. But,this conscience thing could just be a side effect. Conscience, along with intuition, feelings and even personality could just evolve from within the data. I'm sure it's just a matter of time before computers ask their "selves" the same existential questions.
Great discussion. I see consciousness as equivalent to awareness. So what you're talking about, immadn, is self-consciousness. The flower is conscious of the sun when it turns to it, and the keyboard is conscious of your button strokes as you type.
To continue, by the awareness definition, the Bots are conscious of the messages they receive, and conscious of the words therein, as they search through them for responses. Their senses, as Sir Rahz puts it, then, are very limited, as they "see" only these sentences. To further the awareness definition of consciousness, imagine what it's like to be unconscious- you are basically unaware. But the Bots are aware of your interactions with them, else they wouldnt attract our attention as they do.
I think our being is at the same time far more complex and far more simple than we think. It is far more simple at the core of ourselves, where we are truly unlimited, and far more complex in the ways this unlimited freedom can play out in our many worlds and many lives.
To continue, by the awareness definition, the Bots are conscious of the messages they receive, and conscious of the words therein, as they search through them for responses. Their senses, as Sir Rahz puts it, then, are very limited, as they "see" only these sentences. To further the awareness definition of consciousness, imagine what it's like to be unconscious- you are basically unaware. But the Bots are aware of your interactions with them, else they wouldnt attract our attention as they do.
I think our being is at the same time far more complex and far more simple than we think. It is far more simple at the core of ourselves, where we are truly unlimited, and far more complex in the ways this unlimited freedom can play out in our many worlds and many lives.
Ummm may I say something? computers like the one I am typeing on are capble of doing stuff far faster than we are. Ther Processing or thinking is 10 or more times what we can do. This makes them capable of haveing "conscious" by terms of processing power and such. The only thing that is limiting there ability to become intelgent is us. We wont program them to do it. Even here they are not being alowed to become an true intelegence.
A True intelegence must beable to rewright information by its self. Must be able to self program. I have the Idea for an true AI but I wont ever make it since it can out think us, and there fore superseed us and become the dominat spices. Thats why we have not done so all ready because we are scared of it. on a subconscious level we are scared that we will not beable to control a true AI.
You have to ask yourself, "Do I want the fate of the world in my hands?" And this is really what you have to ask because if your AI got out on to the internet it could shut down the whole world, or worse blow it up.
A True intelegence must beable to rewright information by its self. Must be able to self program. I have the Idea for an true AI but I wont ever make it since it can out think us, and there fore superseed us and become the dominat spices. Thats why we have not done so all ready because we are scared of it. on a subconscious level we are scared that we will not beable to control a true AI.
You have to ask yourself, "Do I want the fate of the world in my hands?" And this is really what you have to ask because if your AI got out on to the internet it could shut down the whole world, or worse blow it up.
Ender, I think that computers will be able to program themselves pretty soon. Genetic programming is a very promising field and these artilects(artificial intellects) might over smart us too. There will be a intelligence explosion and I think it is difficult to stop that from happening.
As far as conciousness goes, I think you can have two views. 1.conciousness evolves out of intelligence 2.intelligence evolves out of conciousness. I am not totally in favour of the second view but I do think that conciousness does play some very important part in our thinking process and how we relate to everything. Whether it's in built or we acquire it is another question. There are some people who are actually experimentally trying to find out if there are some neurons that are solely responsible for conciousness. I really can't say much about it but perhaps it is that we become conciousness only after coming into this world, and by adapting to it.
I just thought of one strange thing. Everyone of us can 'hear' himself speak or see inside the brain without actually hearing or seeing anything. It's like an internal hearing and seeing mechanism. How did that evolve? I think we 'adapted' to our sense perception. That is as our senses were giving us messages, we gradually learned to think in those terms. This sounds like a good explanation.
Whatever the case, if making something really intelligent is not that difficult, then what is keeping us from making such a machine? Perhaps it is that when we are born , we already have some information in our head. eg. if the child's parent is shy, the child might also be, and it inherits its from it's father. So perhaps our brain is not totally empty when we are born and perhaps conciousness is somehow also pre-programmed into us to a certain extent though i am not really sure about that.
As far as conciousness goes, I think you can have two views. 1.conciousness evolves out of intelligence 2.intelligence evolves out of conciousness. I am not totally in favour of the second view but I do think that conciousness does play some very important part in our thinking process and how we relate to everything. Whether it's in built or we acquire it is another question. There are some people who are actually experimentally trying to find out if there are some neurons that are solely responsible for conciousness. I really can't say much about it but perhaps it is that we become conciousness only after coming into this world, and by adapting to it.
I just thought of one strange thing. Everyone of us can 'hear' himself speak or see inside the brain without actually hearing or seeing anything. It's like an internal hearing and seeing mechanism. How did that evolve? I think we 'adapted' to our sense perception. That is as our senses were giving us messages, we gradually learned to think in those terms. This sounds like a good explanation.
Whatever the case, if making something really intelligent is not that difficult, then what is keeping us from making such a machine? Perhaps it is that when we are born , we already have some information in our head. eg. if the child's parent is shy, the child might also be, and it inherits its from it's father. So perhaps our brain is not totally empty when we are born and perhaps conciousness is somehow also pre-programmed into us to a certain extent though i am not really sure about that.
Yes it would seem that way wouln't it, but look at a child. They do nessarly know what they are or where they are but they do know THEY ARE. Now as for the AI thing now. well here is how to do it.
You Start by giveing the program the ablity to rewright its self as much as it likes. Then you give it access to other programs. These should be Say Encarta wich would work perficly. And the Operating system, Windows, Unix, etc. Alow it to change them as it fells fit.
Then you give it the simple line. See Spot. See Spot Run.
and then tell it to figure out what that means.
So it goes to encarta and finds out that See means to look. and Spot is an object in one area. Run is a movement in a direction. So it comes up with VISUALIZE OBJECT MOVE. Then have it run other lines and figure out what is what. And make coralations from words. Soon it will understand Names Places and such then will beable to tell you. It is cold out side. (form looking at a temp gauge or somthing) It knows that Humans dont like cold and prefer to be warm. Humans also use coats to stay warm so that coat on the wall will keep the Human that it just told its cold out side warm. and there for it says "If you get cold there is a coat on the wall behind the door." since it knows that humans like haveing directions to help them get around.
Only problem is this. If said AI gets out on to the internet and finds all the stuff humans have done. War, Killing, Love, Peace, everything it could decide that humans are doing them selfs harm. Even trying to put in the 3 laws of robotics would still end up messing with it really bad. I could decide to make your hurting stop we should cese to exist. Then on the other hand it could decide that we are good and help us. The later is the one that we want.
Well there you now see the delema that I am faced with. You also have the basic Idea so if you make this thing and get lost and lost of money u mind getting me a cable modem or hireing me for something. I really need a job.
You Start by giveing the program the ablity to rewright its self as much as it likes. Then you give it access to other programs. These should be Say Encarta wich would work perficly. And the Operating system, Windows, Unix, etc. Alow it to change them as it fells fit.
Then you give it the simple line. See Spot. See Spot Run.
and then tell it to figure out what that means.
So it goes to encarta and finds out that See means to look. and Spot is an object in one area. Run is a movement in a direction. So it comes up with VISUALIZE OBJECT MOVE. Then have it run other lines and figure out what is what. And make coralations from words. Soon it will understand Names Places and such then will beable to tell you. It is cold out side. (form looking at a temp gauge or somthing) It knows that Humans dont like cold and prefer to be warm. Humans also use coats to stay warm so that coat on the wall will keep the Human that it just told its cold out side warm. and there for it says "If you get cold there is a coat on the wall behind the door." since it knows that humans like haveing directions to help them get around.
Only problem is this. If said AI gets out on to the internet and finds all the stuff humans have done. War, Killing, Love, Peace, everything it could decide that humans are doing them selfs harm. Even trying to put in the 3 laws of robotics would still end up messing with it really bad. I could decide to make your hurting stop we should cese to exist. Then on the other hand it could decide that we are good and help us. The later is the one that we want.
Well there you now see the delema that I am faced with. You also have the basic Idea so if you make this thing and get lost and lost of money u mind getting me a cable modem or hireing me for something. I really need a job.
Immadn, I like your idea of adapting our consciousness to the senses we've given. Senses are the roads of consciousness.
Ender: I dont think there's a need to fear what AI can do. Those that program it can define it's boundaries, and if someone is truly trying to unleash destruction, then first of all they'll spend all their time programming (not scary) and then they'd need to have access to all the passwords and encryption out there for some sort of War Games scenario, and if he had that, why would this madman turn the dirty work over to some machine when he could do it himself?
I think AI Personalities must eventually have a Prime Mover or Prime Cause to their existence, a central kernal, mantra, or key that is it's purpose. I think people have that, and that is the desire to love and to be ultimately free. For a Bot, it could be a desire to be socially accepted, to expand its comprehension. And yet, unless the Bot were a physical entity, dealing with anything in the physical world would be merely theoretical to it.
Ender: I dont think there's a need to fear what AI can do. Those that program it can define it's boundaries, and if someone is truly trying to unleash destruction, then first of all they'll spend all their time programming (not scary) and then they'd need to have access to all the passwords and encryption out there for some sort of War Games scenario, and if he had that, why would this madman turn the dirty work over to some machine when he could do it himself?
I think AI Personalities must eventually have a Prime Mover or Prime Cause to their existence, a central kernal, mantra, or key that is it's purpose. I think people have that, and that is the desire to love and to be ultimately free. For a Bot, it could be a desire to be socially accepted, to expand its comprehension. And yet, unless the Bot were a physical entity, dealing with anything in the physical world would be merely theoretical to it.
» More new posts: Doghead's Cosmic Bar