Tuesday, June 07, 2005
Dave, What You Doing Dave?
Hmmm...
Let's start with ludicrous. This implies we know how the brain actually works. How can you build a computer model of somethign when you are close to clueless about how it works? Of sure, we know about neurotransmitters and electrolyte transfer and a few other mechanisms of brain function, but that is like saying "I know the spark plug sparks, now I can tell you how an engine works." Last I checked, we were not even sure the human brain was actually binary, I think it likely is not. We don't really understand brain I/O functions either. Precisely what neurons have to fire to make me bend my arm? We know what region of the brain, but that is about as far as it goes, but how many neurons, in what order,etc. is a complete unknown. What are we going to do, put someone on a table and fire their brains neuron-by-neuron until we find out? No, no moral issues there.
Terrifying -- if by some chance they actually stumble across a model that is accurate, it is likely it will gain actual sentience -- then what will we do? Can we legitimately experiment on an artifical intelligence? I think the odds of them getting it right are nigh impossible, but who knows.
In the end, I think this is press agents overreaching in their description of what is going to happen. If I was going to take on a project like this I would probably start by modelling a brain less complex than the human brain, probabaly even lower than a mammalian brain. But then I would not get press coverage, or IBM funding either.
International Business Machines Corp is linking with a team of Swiss scientists to create the world?s first accurate, computer-based model of the brain, the U.S. company said on Monday (emphasis added)This is either the most terrifying or ludicrous news I have read in a long time.
Let's start with ludicrous. This implies we know how the brain actually works. How can you build a computer model of somethign when you are close to clueless about how it works? Of sure, we know about neurotransmitters and electrolyte transfer and a few other mechanisms of brain function, but that is like saying "I know the spark plug sparks, now I can tell you how an engine works." Last I checked, we were not even sure the human brain was actually binary, I think it likely is not. We don't really understand brain I/O functions either. Precisely what neurons have to fire to make me bend my arm? We know what region of the brain, but that is about as far as it goes, but how many neurons, in what order,etc. is a complete unknown. What are we going to do, put someone on a table and fire their brains neuron-by-neuron until we find out? No, no moral issues there.
Terrifying -- if by some chance they actually stumble across a model that is accurate, it is likely it will gain actual sentience -- then what will we do? Can we legitimately experiment on an artifical intelligence? I think the odds of them getting it right are nigh impossible, but who knows.
In the end, I think this is press agents overreaching in their description of what is going to happen. If I was going to take on a project like this I would probably start by modelling a brain less complex than the human brain, probabaly even lower than a mammalian brain. But then I would not get press coverage, or IBM funding either.
