If the purpose is to understand the intelligence we should look at the only example we have at hand: the brain. We should ask ourselves how does the brain gives room for intelligence and consciousness.
Yet to know you've understood how it works, you need to experiment. Either playing with a living mammal brain, or testing your hypothesis in simulation. Regardless of how much I'd like to do the first, that's out of the table and so, simulations will have to do. Fortunately we are computer scientists and that's something we can somehow do.
The easy answer then is to simulate the brain and the tinker with it until everything is understood. But... can the brain be simulated?
Short answer: no.
Not so short answer: not yet.
Long answer: Simulating stuff is not easy. People who works doing simulations usually reduce the thing they want to simulate to the basics relevant to the simulation. If the simulation wants to show how an object moves, regarding the gravity, then the object is reduced to a position, weight and its interaction with gravity. It's irrelevant what's its color or what kind of atoms it is composed of.
However we don't know the basic properties and rules in the brain that are responsible for the existence of intelligence, meaning that regardless of how much we know about the brain, we still don't have an answer to "why from all the things that compose the brain intelligence emerges?"... So if we want to simulate the brain, we'd have to simulate it all (each neuron, each neurotransmitter, each tiny element of it, plus their interactions). To achieve so we'd need a really big interdisciplinary team working without rest for years, and what's worse, the computational power of a supercomputer that Google would dream to have. We have neither.
Good news is: simulating a brain means having created artificial intelligence.
So we're not talking of sentience here, that's a horse of a different color. We're talking about properties that can be actually be measured, but that's material for the following posts. We only need to say this: if it displays intelligence, then it's intelligent.
So we are in a quest to reduce/simplify the brain so that we can simulate it.
Once simulated, our quest is completed. Why? Because It means we've found the basic properties that are responsible of the existence of intelligence and now we can use it to create artificially intelligent agents. We'll leave the philosophical discussion for another moment. (Read hard problem of consciousness)
Yet to know you've understood how it works, you need to experiment. Either playing with a living mammal brain, or testing your hypothesis in simulation. Regardless of how much I'd like to do the first, that's out of the table and so, simulations will have to do. Fortunately we are computer scientists and that's something we can somehow do.
The easy answer then is to simulate the brain and the tinker with it until everything is understood. But... can the brain be simulated?
Short answer: no.
Not so short answer: not yet.
Long answer: Simulating stuff is not easy. People who works doing simulations usually reduce the thing they want to simulate to the basics relevant to the simulation. If the simulation wants to show how an object moves, regarding the gravity, then the object is reduced to a position, weight and its interaction with gravity. It's irrelevant what's its color or what kind of atoms it is composed of.
However we don't know the basic properties and rules in the brain that are responsible for the existence of intelligence, meaning that regardless of how much we know about the brain, we still don't have an answer to "why from all the things that compose the brain intelligence emerges?"... So if we want to simulate the brain, we'd have to simulate it all (each neuron, each neurotransmitter, each tiny element of it, plus their interactions). To achieve so we'd need a really big interdisciplinary team working without rest for years, and what's worse, the computational power of a supercomputer that Google would dream to have. We have neither.
What's else then?
Nothing else, sadly.We can only do what simulators do, and try to find a simplification of the brain (a reduction of the brain) and test if it does what it is supposed to do.Good news is: simulating a brain means having created artificial intelligence.
So we're not talking of sentience here, that's a horse of a different color. We're talking about properties that can be actually be measured, but that's material for the following posts. We only need to say this: if it displays intelligence, then it's intelligent.
So we are in a quest to reduce/simplify the brain so that we can simulate it.
Once simulated, our quest is completed. Why? Because It means we've found the basic properties that are responsible of the existence of intelligence and now we can use it to create artificially intelligent agents. We'll leave the philosophical discussion for another moment. (Read hard problem of consciousness)
Comments
Post a Comment