Reply to thread

It would be good, the problem with computers atm is that when you get right down to it, it's just a fancy calculater.

Take your example. "Computer how do I cure cancer"

Now if programmed right a powerful enough computer could come up with the most efficient treatment for example we now have several options in pill form that fight cancer.

In this case the computer knows from a database which chemical compounds are used in cancer drugs and simulate different combinations of them, for the most stable and effective model.

But let's say the tests on cannabis actually do cure cancer. A computer could never even think to use cannabis in its modelling because classification would be listed as illegal narcotic medical use: glaucoma, pain relief.

The computer can never think of anything new, the computer could find the best result of existing treatments because all its doing is simulating the result of thousands upon thousands of drug combinations in effect it's shortening the work humans have to do to find stable effective treatments.

Now don't get me wrong this would speed up research dramatically. But the computer has no understanding of things, it can not actually learn, only run calculations.

And that is why the computer cannot find something that humans wouldn't have thought of, because it relies on human input for everything.

The problem we have is this concept that the computer can have information we don't or figure things out that we can't.

This is a myth, take for example deep blue the computer that ruined chess. Now all deep blue does is calculate every possible action that the opponent can take and as a result chooses the most correct move possible.

A sound feat, until you realise known of this is learned, it's all programed a human sat down and inputted every single possible chess move, they programed the rules, they programed board, the pieces everything.

All the computer does is take that information and run a bunch of chess games virtual to find the move that is the best case of that scenario.

Deep blue ran something like 200 million moves per second, even with all that power the computer couldn't win every game, this is because the computer doesn't actually think. It just calculates. It should also be noted that between games a human actually sat reprogrammed deep blue to adjust for kaspeirvos(sp) moves, because the computer would have found his moves as insignificant in terms of statistical likelihood in unorthodox moves resulting in a win, a key reason why the machine lost the previous year.

There has even been speculation that IBM even cheated to achieve victory in 97 suggesting that people were making adjustments mid match.