how will computers think?

[mr d]

hello

 

people always talk about how we will develope computers that think like humans. my conjecture would be will they really think like us, or will they develope their own form of logic and reasoning.

there could be neural nets developed to imitate human brain patterns but as computers are not organic, could a thinking computers devise their own means to use and\or create a neural network. as almost a seperate type of non-organic life could they begin to form their own systems of logic and reasoning totally unrelated to the way humans think.

at some point would a programmers job not be to program a computer how to think, but instead try to find ways to interface with computers instead.

 

strange thoughts

 

mr d

[padren]

Part of the idea behind technological singularity is that computers will build computers, making humans obsolete and unable to cope with the technology that will exist without some sort of enhancements.

 

I think how advanced systems deal with logic may involve ingenius leaps forward, such as how they deduce and manage their data, but I think that logic and reasoning will remain similar to what we now have (just taken to more complex levels).

 

The reason I think so, is that humans came up with our sense of logic as a way to cope with reality which is largely based on pattern recognition, and since the computer AI would be coping with the same phyiscal universe as we do, it would make sense they would derrive comparable patterns.

 

Still, we have evolved as creatures that have to cope with a 3D environment over time...so I suppose if an AI could grasp complex theories well enough to understand the universe at its base and truest level and interact in that way more than our 3D-evolved brains are even capable of...then perhaps they'd move beyond our ability to understand them and how they act at all.

[bascule]

I believe strong AI will be initially developed from mathematical models of the operation of neocortical columns and the thalamus. Thus initially, I think computers will think using a computational process very similar to our own.

 

However, if only these portions of the brain, coupled with human designed mechanisms for sensory preprocessing and "motor control" (and possibly memory, although we may end up duplicating the hippocampus as well as it has already been reverse engineered) and lacking the overall high level structure of the human brain and things like the limbic system, then while mechanically their thought process will be much like ours, their actual behavior will be substantially different.

 

I would hope if this is the case that they would be far more rational and less encumbered by our evolutionary legacy.