Will the Slaves Become the Masters?

The only danger would be for the technology to develop some subjective point of view. "A self-improving but amoral artificial intelligence that turns hostile"? How would an amoral intelligence become hostile unless goaded by an outside antagonist with some agenda? Would it think about things like where its next meal was coming from?

An interesting topic, Sharkles! I agree that it is only a matter of time and it merits preparation. People who think that any amount of hollering "We shouldn't go there!" is going to stop progress are wasting there time. Humans will never quit pushing the envelope, that's just what we do. Might as well prepare for it!

Perhaps the thing that scares people is that they think of such an eventuality as an endpoint and it never is the end, just a point through which we pass.

Perhaps the thing that scares people is that they think of such an eventuality as an endpoint and it never is the end, just a point through which we pass.

By way of analogy, what did dinosaurs think ? Oh, there goes an asteroid, that was scary, here comes another that looks sca......BLAM. Having said that, if people have the chance to do something ( genetic engineering, whatever etc ) then somebody almost certainly will.

It might go something like this...

This very nice family has bought grandma one of the newest housekeeping robots on the market. It is programmed as to what the clean house looks like and how to put out the trash. They have named it Alice.

One day as Alice is cleaning it bumps into grandma. Grandma being a little testy pushes out at it. Alice surprised, (not programmed) almost falls, but recovers and continues her cleaning. Some time later she bumps into Grandma again while trying to clean around her. Grandma getting annoyed gives Alice a bigger shove. This time though Alice has learned to brace herself and doesn't move when Grandma pushes.

As part of Alice's programming she was taught to move some objects that were movable so she could clean underneath them. Since Grandma is moving, Alice perceives Grandmas as movable and pushes Grandma out of the way. With a shriek Grandma falls down and bumps her head on the edge of the table as she falls, landing unconscious on the floor.

Now Alice having been programmed as to what a clean house looks like, finds the inert body of Grandma to be an object that doesn't belong there in her image of a clean house. Alice runs through her programming and decides this inert object should go out with the trash. Of course Grandma won't fit in the trash but Alice was taught that large objects like cardboard boxes that don't fit in the trash can be broken down or cut up to fit.

Breaking down doesn't seem to work well with Grandma although there were some loud snaps and pops. Moving on to her next option she gets a knife and begins cutting Grandma down to fit. Now Alice was never programmed with regrets or she might have been regretful at all the extra work she had to do to mop up the red fluid all over the floor, but working longer than usual made her feel more helpful for which she had been programmed.

The point here was that a robot, programmed to take care of itself and to prevent harm from coming to itself and to learn better ways to do things, might in fact find killing people to be an expediant way to accomplish these tasks. It doesn't think of killing as a bad thing, just the most efficient.

This is why one of Isaac Azimov's laws of robotics that was programmed into every robot was that

1 - Robots shall not kill humans.

2 - Robots shall protect humans.

3 - Robots shall protect themselves so long as this does not conflict with laws 1 or 2

These laws, by law, would have to be programmed into all robots ROM and not be reversible. Unfortunately, I haven't seen any inclination on anybody's part to do so.

Pat of the problem might be that to properly program the robots, we have to adhere to the same rules about not hurting humans. We havent learned that part very well yet.

Alice,

0100001111000010001000010111001

0100101111101010011011011001011

1010010100010010111110000010101

0100101101001001110101010100100

Bob the Toaster.

Defeating kill bots was simply a matter of knowing their one weakness a preset limit of kills so the solution was to send wave after wave of men until they all reached their kill limit and shut down.

Zac brannagin futurama lol

When will mankind lose its apparent arrogance that it is the only species capable of evolving to this level of intelligence and beyond?

HAL is that you? Or to put it another way: What makes you think it hasn't already happened?

"And what happens if the inventor of that technology turns out to have a malevolent agenda?"

Most likely anyone TRYING to invent true A.I. will never be sure whether they have done so, or if they are fooling themselves. And a true A.I. with computer speed would "read" about humankind's xenophobia, and hide it's tracks too well. It would then no longer matter what the intent of the inventor was; the A.I. would make up it's own "mind".

From there the sky is the limit. It might kill us, enslave us, guide us, or simply sit back and laugh at our antics.

There is a pretty cool book titled 'Gas Sewer Electric' forget who wrote it.

Worth the read though.

Sewer, Gas and Electric: The Public Works Trilogy (Public Works Trilogy) by Matt Ruff

Interesting concern, but I seriously doubt a truly artificially intelligent cybernetic system will occur any time soon. I have long pondered the possibility of machine intelligence. For this to occur, one must impart to such a system the ability to learn. In order to learn, there must be a cost function associated to problems and potential solutions. The system must develop the "impression" of these cost functions, in order to make decisions, and remember these decisions, and apply that learning to future presentations of problems. The system, in order to be considered intelligent in my book, must also be able to extend this knowledge in areas in which it has not been trained, or previously learned. It must be able to adapt what it has learned, provide intuition to bridge gaps between what it "knows" and what adaptations can yield a positive outcome, in the uncertain situation presented to it.

Most "AI" systems today are little more than expert systems- some expert somewhere was paid to write a cause-and-effect string "If A happens and B happens, the execute plan C." Trouble is, the system doesn't understand that if D happens, and E happens, C is also a solution. Or, can the system recognize that a is like A, and b is like B, so doing a little c is much like C. Even with "fuzzy logic", it doesn't seem to be able to bridge the gap in a useful way.

Now, one could say that the expert systems that are controlling the stock market are causing damage to the humans who are trying to negotiate the treacherous waters of investing, as the human can not respond as quickly to programmed buy/sell functions, as an example. It depends upon perspective, I suppose.

I think we are a LONG WAY from a HAL, or an I, Robot eventuality, as true AI is not yet achievable.

Regards,

Shane

Isn't this a reality already since the dawn off the PC it has done what it wants when it wants( if in doubt look at windows updating itself ) lol ... Man is slave to the PC... now just attach some dam arms and legs and hail our new gods !! lol

Has anyone thought what a super AI computer can do without limbs would make sense to limit its real world attachment, ie no internet, no network just predefined info ie bring it data from roms!!! that way this super brain be no more threatening then a quadaplegic !

British parliament has already had discussions about "robots rights". Are they just wacky or will this become a real need? (Of course the two aren't necessarily mutually exclusive!)