Something to Consider

[ARefinedGent]

Hi, I have a rather strange question, but there's probably no better place to ask it than here. I was recently replaying the classic Nintendo series "Mega Man" (for those ignorant to video game, the premise is rather simple: a "good" scientist named Dr. Light creates various robots for everyday tasks - Metal Man, Air Man, Snake Man, etc., including his prized robo, the titular character Mega Man; but then an "evil" scientist named Dr. Wily steals and perverts these robots for his own diabolical use). Anyway, my question is this: if there comes a day when robots walk among us, and are not only used but depended upon for our comfort and safety, what methods can be put in place to stop the Dr. Wily's of the world from reprogramming them to attack us? 

[paulstreats]

In short, nothing.

 Its all really down to programmed code architectures. We are very far off creating a system where robots can serve us reliably in an everyday social type capacity. It is only expected though that it will take some mad or evil person just as long to produce his/her own architectures to replace the ones originally intended(and probably working on their own it could take a life time)

One thing that i always think of when questions of robot ethics are raised is that robots that are meant to do good can eventually turn bad once we get to a level of programming architecture to make this kind of thing possible. On the other end of the scale, some humans turn out out doing bad no matter how they are brought up. A person can be brainwashed into suicide bombing and so can a robot, so wheres the difference? whats the problem?

[ed1380]

exactly. if you hardwire the robot for good, the hardware can be easily changed for evil.
that's what emp nades are for

[SciOlyStudent]

Wow, what a classic question.  I believe that if we get to fully autonomous robots with full functioning A.I., we would never learn how to put limits on it.  Think of this, a robot brain has access to the internet and you programmed its mind to "learn".  Just by reading, say, a suicidal website, the robot would "learn" that information and try to kill itself.  No matter how advanced of a robot mind we make, we will never really be able to make it discern good from bad, because to it information would be information.  The only way we could keep "bad information" from corrupting what we want to be a good robot, we would need to specify every last little thing that we deem as bad.

[TrickyNekro]

I indeed think think that absolute AI will be never achievable as me mean it... Even if they learn, even if they have access to the Internet... even if what ever you might think... They STILL have to choose... choices can be preprogrammed... but not infinitely... Even supposing we have a enormous database of reactions... STILL they are limited... Machines themselves will never act on their own.... They just react...
My point is that AI is not possible with sand and metal... just automatic machines...
But that's what biotechnology is for!!! Why do you think they are trying so hard to have nerves grow up alone on IC boards!!! (they have actually done in France if I'm correct...)
What is our brain after all... Nerves... (I hope you don't mind that I do not believe in god - not just a teen reaction but through "practical" thinking!!!)
Of course we may not see all these... Either our children... (If we have children and the planet isn't dead by then...)
After all robots can be whatever we want 'em to be!!!
---->http://rapidshare.com/files/77822079/The_truth_plain_and_simple_.rar.html
Not talking about robots but still nice having a look!!!

Cheers, Lefteris

[frank26080115]

Ah, you have to think about programming differently. It could not be purely digital and binary. There is research being done using clusters of computers to emulate the known behaviors of animal brain cells. If a successful version of that isn't going to be considered AI, then I don't know what it is.

[SciOlyStudent]

It may be considered A.I., but what could it do?  There would be no way (except through programming as we know it) to make it interact with anything.  And lest we forget, computers are programmed using digital and binary programming, and if it is computers that are making these groups of cells, that A.I. is still programmed based, limiting it to what we put in it.

[Kohanbash]

you should read "How to survive a robot uprising"