A few of us regularly have these arguments (which of course ends up with physical damage to various human bodies), about artificial intelligence. As to whether any computer will ever beat the “Turing test”. Or will that be an insatiable dream.
My friends feel that the computer will never be able to decipher patterns which it has never previously seen. (What I mean here is if a computer is given a sequence, and is expected to predict the next item in the sequence, if the computer has never seen the pattern it will fail.)
But here I insist that humans also work on a similar pattern recognition algo. So if you see a pattern which you have not seen, you will ask your friend and learn. And this learning is what I feel is the actual problem. A kid for example, does not understand anything when it is born. Then the parents/surroundings teach it. It falls, learns from the mistake and the next time it comes across the same scenario it takes a different course of action.
Does the above not sound similar to a chess game (with learning abilities) or google with the human click as a factor for deciding the relevancy?
I feel it is similar. Yes, there are some problems:
1. The computer does not understand emotion – this is something I still am thinking about L
2. The computer can only learn patterns which it has been coded to learn. So if a programmer does not code a particular learning pattern, the computer may miss out on some classes ;)
- This I feel is a problem which is more related programming languages than to the computer itself. And I feel the answer to this may be in Dynamic Languages instead of the current Static Language
3. Humans do not interact with the computer enough –
- A large part of our interaction is not done over computers. For example, you attend a meeting where some sleep while others give valuable information, you call up your friends and share some information over the phone, you read a newspaper (a valuable source of your tastes) and so on. None of these involve a computer. So the computer misses out on the learning curve!!!!
- This I feel is one among the biggest problems why a computer is unable to replicate a human being.
- So to build up some AI, what we actually need is a single location thru which all information flows. Of course some persons get itchy about this thought, but currently I assume ideal working conditions
To be concluded in part 2.
Comments