Robots vs. Human
Despite many fictions and movies and much of our own imagination, there will never be a war between robots and human. If that war ever get started, it will be over before one would realize. Computers will never reach the sophistication of a human mind. Computers will never survive an uncontrolled environment.
We take it for granted, and we never appreciate how magnificent our mind is -- until we trying to create one with computers.
Computers are really good at following specific, extremely simple instructions, such as add a number, output 5 volts signal, etc. They can do it very fast, faster than any human mind can do. The reason behind computer's this ability is because it does not think. It does not think which next instruction should be, how valid next instruction is, and consequence of executing an instruction. If you give it an instruction to shutdown, it does it without "hesitation". It is like drive a car at full speed without breaks and wheels. The reason it doesn't crash is only because it runs at a controlled environment. It works because we "think" for them and created the empty space for them to run headlessly.
Human mind is the result of evolution, a result to survive an uncontrollable, largely unknown world. Flexibility is the key to survive. In order to survive, we need be able to change our action before it is too late. If we are running fast, we need be able to stop or turn direction before we hit the wall. This flexibility significantly slows us down, but it is necessary for us to survive.
It is extrememly sophisticated for our mind to possess this flexibility. Lets say you were running forward, then you need prepared to stop is a wall showing up in the front. Then you need prepared to recognize the wall when you see it, or hear it, or sense it with your sixth sense. Then you need process all thoses signals coming from your senses, determing which signal is OK to ignore, and which signal needs more process, and which signal need necessary change of action. If we are programing the robot to do all these, we are writing "if"s and "else if"s pages after pages. There are simply too many variables out there.
If all the variables are known, the robot still can win. It is just a matter of ennumerate all the variables and choose corresponding actions on any of the cobination of these variables. Surely it will run much slower than reckless running, but given the exponential growth of cpu speed, it may still run faster than human after all these "thinking". Then, there is one last necessary part in order to survive -- prepare for the unknown. We need the flexibility of taking necessary actions to survive if an unknown variable or a variable with unexpected values pops up. Should we proceed? What if that "unknown" is deadly to proceed? Should we turn or even return? What if that "unknown" is nothing serious? Most robots will either crash or unnecessarily stop functioning at this point.
Well the key that human deal with unknown is risk judgement and learning. To judge the risks, we are able to quickly connect the unknown to what is known, not by enumerating all the memory we have, but by massive parallel retrieving the most relavant memory, and quickly reach a judgement. Well, frankly, I don't know I our mind does it. I just appreciats that it does it. Now comes learning, which involves a feedback on the result or concequence of our action against that "unknown". Again, another amazing process.
So, after postulate on all these amazing stuffs our mind does, I don't think it is slow at all. If you program a database, you may have that experience of slowness when retrieving certain data over a large database. Well, human mind can retrieve the most relavant info on a split second. Do you ever have that feeling of instincts? Instincts shows up in mind faster than we can reallize, which essentially is a data retrieving process.
Well, I should not say definitely that robot will not evolve to the sophistication that matches human mind. But I suspect, but reaching that sophistication, the computer probably will be bulky and slow, and thus, useless.
We take it for granted, and we never appreciate how magnificent our mind is -- until we trying to create one with computers.
Computers are really good at following specific, extremely simple instructions, such as add a number, output 5 volts signal, etc. They can do it very fast, faster than any human mind can do. The reason behind computer's this ability is because it does not think. It does not think which next instruction should be, how valid next instruction is, and consequence of executing an instruction. If you give it an instruction to shutdown, it does it without "hesitation". It is like drive a car at full speed without breaks and wheels. The reason it doesn't crash is only because it runs at a controlled environment. It works because we "think" for them and created the empty space for them to run headlessly.
Human mind is the result of evolution, a result to survive an uncontrollable, largely unknown world. Flexibility is the key to survive. In order to survive, we need be able to change our action before it is too late. If we are running fast, we need be able to stop or turn direction before we hit the wall. This flexibility significantly slows us down, but it is necessary for us to survive.
It is extrememly sophisticated for our mind to possess this flexibility. Lets say you were running forward, then you need prepared to stop is a wall showing up in the front. Then you need prepared to recognize the wall when you see it, or hear it, or sense it with your sixth sense. Then you need process all thoses signals coming from your senses, determing which signal is OK to ignore, and which signal needs more process, and which signal need necessary change of action. If we are programing the robot to do all these, we are writing "if"s and "else if"s pages after pages. There are simply too many variables out there.
If all the variables are known, the robot still can win. It is just a matter of ennumerate all the variables and choose corresponding actions on any of the cobination of these variables. Surely it will run much slower than reckless running, but given the exponential growth of cpu speed, it may still run faster than human after all these "thinking". Then, there is one last necessary part in order to survive -- prepare for the unknown. We need the flexibility of taking necessary actions to survive if an unknown variable or a variable with unexpected values pops up. Should we proceed? What if that "unknown" is deadly to proceed? Should we turn or even return? What if that "unknown" is nothing serious? Most robots will either crash or unnecessarily stop functioning at this point.
Well the key that human deal with unknown is risk judgement and learning. To judge the risks, we are able to quickly connect the unknown to what is known, not by enumerating all the memory we have, but by massive parallel retrieving the most relavant memory, and quickly reach a judgement. Well, frankly, I don't know I our mind does it. I just appreciats that it does it. Now comes learning, which involves a feedback on the result or concequence of our action against that "unknown". Again, another amazing process.
So, after postulate on all these amazing stuffs our mind does, I don't think it is slow at all. If you program a database, you may have that experience of slowness when retrieving certain data over a large database. Well, human mind can retrieve the most relavant info on a split second. Do you ever have that feeling of instincts? Instincts shows up in mind faster than we can reallize, which essentially is a data retrieving process.
Well, I should not say definitely that robot will not evolve to the sophistication that matches human mind. But I suspect, but reaching that sophistication, the computer probably will be bulky and slow, and thus, useless.
posted by Hui Zhou at 6:30 AM
0 Comments:
Post a Comment
<< Home