12-06-2012, 02:33 PM
build a "fake" world and then play as the dragon in that world and teach the AI step by step. Then seamlessly migrate it from a dragon as NPC in a non real world to dragon as real in a world it thinks is still the virtual world.
I'll "harvest" the digital dragon and move it to a actual dragon body.
Now as to how to make a functional complex game AI. Thats a topic worthy of its own universities and there are many.
The idea is to have it have only two priorities, Find water, Find sunlight. The rest is spent in trying to learn, find and utilized patterns in the surrounding and find alternate sources of Water and Sunlight and then gain access to said sources of Sunlight and Water.
It will start by simply walking to a river and drinking.
then it will walk to a river, notice there is a fence around, stop, look for a way around the fence, if none is found it will avoid going in, if it sees someone in its path but there is a hole in the fence it will engage its speech system and act as if the person in front of the gate has a quest for it before it can pass.
"Hail stranger, i wish to partake of your water and sunlight, may i enter. In return, i can engage in many tasks for you that do not require me to endanger a another self"
Then it uses its massive brain (which would be depending on its size and how well I can optimize the interior, something like 50-200 cores) to engage its human conversation simulator, similar to Siri (that talking phone thing) and if it cant make sense of what is being told it will apologize for the misunderstanding and walk away.
In reality, once the AI is established it will grow almost seamlessly forever. The problem is getting it to work for the start. That i anticipate to be a major bulk of the developement cost but still to be under a half a million based on my estimates and various friends in super high tech related jobs in the cutting edge and their estimates of how much it would cost to build.
It is essentially a program that seeks water and sunlight and new ways to get water and sunlight and while not actively seeking those, it will observe its surroundings and attempt to ascertain patterns such as "the stars appear to move at 0.1% faster today than yesterday, it seems there is a pattern here, this is interesting, catalouge for later reference"
Once it has a massive amount of noticed repeatable patterns it will try to predict based on those patterns and evolve from there. Eventually it should see humans as pieces of the architechture of its soul and not as separates selves.
I'll "harvest" the digital dragon and move it to a actual dragon body.
Now as to how to make a functional complex game AI. Thats a topic worthy of its own universities and there are many.
The idea is to have it have only two priorities, Find water, Find sunlight. The rest is spent in trying to learn, find and utilized patterns in the surrounding and find alternate sources of Water and Sunlight and then gain access to said sources of Sunlight and Water.
It will start by simply walking to a river and drinking.
then it will walk to a river, notice there is a fence around, stop, look for a way around the fence, if none is found it will avoid going in, if it sees someone in its path but there is a hole in the fence it will engage its speech system and act as if the person in front of the gate has a quest for it before it can pass.
"Hail stranger, i wish to partake of your water and sunlight, may i enter. In return, i can engage in many tasks for you that do not require me to endanger a another self"
Then it uses its massive brain (which would be depending on its size and how well I can optimize the interior, something like 50-200 cores) to engage its human conversation simulator, similar to Siri (that talking phone thing) and if it cant make sense of what is being told it will apologize for the misunderstanding and walk away.
In reality, once the AI is established it will grow almost seamlessly forever. The problem is getting it to work for the start. That i anticipate to be a major bulk of the developement cost but still to be under a half a million based on my estimates and various friends in super high tech related jobs in the cutting edge and their estimates of how much it would cost to build.
It is essentially a program that seeks water and sunlight and new ways to get water and sunlight and while not actively seeking those, it will observe its surroundings and attempt to ascertain patterns such as "the stars appear to move at 0.1% faster today than yesterday, it seems there is a pattern here, this is interesting, catalouge for later reference"
Once it has a massive amount of noticed repeatable patterns it will try to predict based on those patterns and evolve from there. Eventually it should see humans as pieces of the architechture of its soul and not as separates selves.