Androids and Gender--Resurrected 11/19!

quote:

Originally posted by Zola Yes, but if you didn't have the laws in the first place, there would be no need for this convuluted reasoning.

You've got to have something like the Three Laws, or something better, like Love. Robots will be tougher, smarter, and faster than humans. If they were uncontrolled, then humans would inevitably be swept aside ala the Terminator movies or worse.

quote:

Originally posted by Zola quote: Originally posted by DoctorZhivago Robots would be likely to consider themselves superior because they'd be faster, smarter, longer-living, and more durable. You can't expect everyone to treat their robots nicely - a lot of people are jerks, and are going to have their superintelligent robots washing their cars. While good treatment would go quite a ways towards avoiding problems, the few problems that would occur would cause mass fear of robots, and the fear would feed upon itself as it caused more confrontations. Faster, maybe, depending on how we built them. Longer living? perhaps. More durable? maybe not, it depends on what we end up doing to make their brains. Smarter? what makes you say that? Seriously, THINK about this for a moment. Culturally, we have a value that says "Robots will be smarter than human beings because they don't carry the emotional baggage that humans do and thus they can use pure reasoning" Just because it's a belief of our culture doesn't mean it is true. I want to look beyond that assumption. I don't agree with it at all. A person who is at one with their emotions is one of the smartest and productive of all, regardless of raw IQ. Pure reasoning is only good for solving certain problems, not all. Sometimes there has to be that leap of intuition that shows you where to look for the solution. You say again and again "Robots will be smarter." I am asking you to examine that statement and tell me WHY you think they are going to be smarter other than having devoured every science fiction book about robots for the last X years. I think androids would probably be very good in some areas and not so good in others. I would think a partnership would work very well. I do agree that some would mistreat a robot if they owned it. In fact, if I was building robots that might become self aware, the only "law" I would try to program something that would permit the robot to refuse orders/stop working if someone mistreated them in the first place.

You raise many good points here. However, robots will be "smarter" than standard humans simply because they will think faster. Their clock rate will go up to the limit of physics, while standard humans are stuck with the electrochemical clock rate they involved with.
I am using the phrase "standard humans" because I think it is likely that most humans will be modified in the not-too-distant future. This will begin to happen when robotic components become clearly superior to natural human ones.
You are already using computers that help you calculate, and communicate farther and faster than you ever could before. You've already taken the first steps. The interfaces will become more powerful and intimate until your computer becomes a part of you. The process will be so gradual that most wont notice it is happening.
At the end of this process, barring some catastrophe, humans will BE the robots, and vice versa. Standard humans will probably live in a small enclave somewhere and be considered quaint. And "pure" robots of the Asimov type, will probably be pretty rare.

quote:

Originally posted by R Trusedale You raise many good points here. However, robots will be "smarter" than standard humans simply because they will think faster. Their clock rate will go up to the limit of physics, while standard humans are stuck with the electrochemical clock rate they involved with. I am using the phrase "standard humans" because I think it is likely that most humans will be modified in the not-too-distant future. This will begin to happen when robotic components become clearly superior to natural human ones. You are already using computers that help you calculate, and communicate farther and faster than you ever could before. You've already taken the first steps. The interfaces will become more powerful and intimate until your computer becomes a part of you. The process will be so gradual that most wont notice it is happening. At the end of this process, barring some catastrophe, humans will BE the robots, and vice versa. Standard humans will probably live in a small enclave somewhere and be considered quaint. And "pure" robots of the Asimov type, will probably be pretty rare.

I think it will take quite a while for them to "think" faster than us on a practical basis because we are patterners and can make that "leap" that connects two seemingly unrelated concepts whereas they would have to go from a-to-b-to-c-to-d

In a couple of thousand years, I agree, though. I suspect by then we will be playing with evolution ourselves and may still be faster in some ways

I agree with you 100% about us incorporating computers into our bodies at some point, in fact, I can tell you in all honesty that for myself, if there was a way right now that I could get a physical link to my computer, I'd do it in a heartbeat because it would be great to have that access without having to be tied to the machine itself.

Imagine not having to carry a calulator because you had it built in...

Shoot shoot, darn darn . . . I've been missing this conversation due to my screwed up internet connection . . . not much time now, but lemmie go through all the stuff I missed and then I'll have some stuff to say (not that I'm the ultimate authority . . . I just want to be included in this).

This is a great thread.

quote:

1.A robot may not injure a human being or, through inaction, allow a human being to come to harm. 2.A robot must obey orders given it by human beings except where such orders would conflict with the First Law. 3.A robot must protect its own existence as long as such protection does not conflict with the First or Second Law. 4.Zeroth Law: A robot may not injure a humanity or, through inaction, allow humanity to come to harm. Modified First Law A robot may not injure a human being or, through inaction, allow a human being to come to harm, except where that would conflict with the Zeroth Law.

quote:

The robot wouldn't take your orders, that would be the Second Law - he would try his best to disarm you. Besides, if you're telling him to kill you, what's to stop it from just quitting when you're no longer a threat? A lot of results would damage/disable the robot, but that's because they were designed never to have to use force - the Laws existed both to keep them from harming humans and to avoid their use as soldiers.

DoctorZhivago:

I still don't see why he wouldn't consider the order despite the second law. If he didn't listen to me, then he'd be breaking the first law because he'd be causing somebody to die by his inaction. Trying to disarm me would be a possibility. However, I was trying to make the example absolutely unambiguous . . . I have the gun up to the hostage's head and I'm ready to pull the trigger (I'm such an evil bastard). One wrong move and BAM the hostage dies. Maybe it would attack me based on the zeroth law . . . since humanity would probably be better off as a whole without me . . .

I still say that the zeroth law is way too ambiguous to really be called a law . . . since it can more or less give an android free will depending on how it interperts a situation and the relative worth to humanity of the various people involved.

However, I readily admit that my Asimov is very very rusty. Fire away . I'd enjoy a discussion about this.

quote:

Originally posted by Zola I think it will take quite a while for them to "think" faster than us on a practical basis because we are patterners and can make that "leap" that connects two seemingly unrelated concepts whereas they would have to go from a-to-b-to-c-to-d In a couple of thousand years, I agree, though. I suspect by then we will be playing with evolution ourselves and may still be faster in some ways

Remember the last Grandmaster versus machine Chess tournament? Big Blue (IBM) won. Simple exhaustive computer search and logic finally beat the best human intuitive Chess master in the world. This happened simply because of speed. And a lot of people had predicted it would never happen. So this transition is occurring right now. Plus, just because no one has yet figured out how to program intuition into a computer, that doesn't mean that someone wont do that soon.

quote:

I agree with you 100% about us incorporating computers into our bodies at some point, in fact, I can tell you in all honesty that for myself, if there was a way right now that I could get a physical link to my computer, I'd do it in a heartbeat because it would be great to have that access without having to be tied to the machine itself. Imagine not having to carry a calulator because you had it built in...

Yes, I think most of us would enjoy that. Besides, as soon as the next computer interface advance is available, economic competition will basically force most people to get the latest one. (Hmm this last part is a little frightening.)

The thing with chess is that it doesn't really have a big range of possible states relative to the real world. There are only so many pieces to keep track of . . . only so many positions each piece can occupy . . . Only one move may be made at a time (no concurrency) . . . Therefore, from any point in the game, a powerful computer like big blue can look ahead at millions of possible states and select the move it likes best based upon that.

But now consider "the wolrd" as our space. To define a state, or a moment of time, in the world at any given time is an overwhelming task indeed . . . I can't even begin to imagine just how much information you'd have to gather. For an android to function in the real world as big blue functions, it would then have start looking ahead through all the world's possible states . . . and that is one hell of a task.

Maybe androids could shrink the space down by only considering things that are so close to them . . . or by only looking ahead so many seconds . . . but these each have very large problems as well . . .

My point . . . an android would not only need speed and capacity to function this way . . . . it would need more speed and capacity than I can possibly imagine.

Now programming intuition would be an answer. That way, an android wouldn't have to plow through all the state analyzing junk. But to program anything, you must have a solid understanding of it, and we don't presently have such an understanding of how our brains work. I guess we're working on it.

quote:

Originally posted by Black Phoenix The thing with chess is that it doesn't really have a big range of possible states relative to the real world. There are only so many pieces to keep track of . . . only so many positions each piece can occupy . . . Only one move may be made at a time (no concurrency) . . . Therefore, from any point in the game, a powerful computer like big blue can look ahead at millions of possible states and select the move it likes best based upon that. But now consider "the wolrd" as our space. To define a state, or a moment of time, in the world at any given time is an overwhelming task indeed . . . I can't even begin to imagine just how much information you'd have to gather. For an android to function in the real world as big blue functions, it would then have start looking ahead through all the world's possible states . . . and that is one hell of a task. Maybe androids could shrink the space down by only considering things that are so close to them . . . or by only looking ahead so many seconds . . . but these each have very large problems as well . . . My point . . . an android would not only need speed and capacity to function this way . . . . it would need more speed and capacity than I can possibly imagine. Now programming intuition would be an answer. That way, an android wouldn't have to plow through all the state analyzing junk. But to program anything, you must have a solid understanding of it, and we don't presently have such an understanding of how our brains work. I guess we're working on it.

Hmm, the number of chess moves on a board is estimated to be 10^120 a pretty large number. Even Big Blue (Hey another Big!) Had to prune its search trees during the chess matches.
Want another example? Computer driving at highways speeds has been successful. You could have a robot car today, except that you couldn't get insurance. (This barrier is now legal instead of technical). This shows a robot can successfully navigate the real world even without having a complete description of it.
Have you ever been surprised? That is simply a mismatch between your brain's "picture" of present/future and reality. It startles you, but it doesn't kill you, and it isn't usually a catastrophe. Sometimes your instincts or reflexes take over when your are surprised. Reflexes are relatively easy to build into robots...

quote:

Hmm, the number of chess moves on a board is estimated to be 10^120 a pretty large number.

I was speaking in relative terms. Yeah, it's a huge number. But the number of possible actions that could happen in the *world* dwarfs it by far. After all, there are probably thousands of chess games going on right now, and they're just a small part of what's happening in the world right now.

quote:

Even Big Blue (Hey another Big!) Had to prune its search trees during the chess matches.

Yeah, I bet. I suppose there are plenty of branches that can just be thrown away outright (sacrificing a queen for no reason, etc). Any real world robot that operated like this would also have to prune the trees . . . obvioulsy nobody is going to keep track of the whole world's state.

quote:

Computer driving at highways speeds has been successful. You could have a robot car today, except that you couldn't get insurance. (This barrier is now legal instead of technical). This shows a robot can successfully navigate the real world even without having a complete description of it.

Curious. I believe you though. The thing I wonder about with this example is how a robot car (for example) would react to something not only unexpected but also unknown. Say, for example, that the robot is watching all of the cars nearby so as not to run into them. However, it is not concerned with the people inside the cars. Then, say that somebody suddenly jumps out of a car and onto the highway. How does the robot car react? Supposing it was not possible to break in time, I guess the car would try to swerve around the person . . . except that the person's moves aren't predictable and maybe they'll get hit.

Or, what if there was a huge fire across the road. The car would survive it, but would the computer understand what fire is and that it's dangerous to its passengers?

I guess what I'm saying is that mistakes become inevitable as you prune the search trees more and more. But that scenario would be exactly the same with a human driver too (well, not the fire one). Still, I imagine that these cars would be less accident prone than manually driven ones, but many people wouldn't trust them . . .

Hope this was halfway comprehensable . . .

quote:

Originally posted by R Trusedale quote: Originally posted by Black Phoenix The thing with chess is that it doesn't really have a big range of possible states relative to the real world. There are only so many pieces to keep track of . . . only so many positions each piece can occupy . . . Only one move may be made at a time (no concurrency) . . . Therefore, from any point in the game, a powerful computer like big blue can look ahead at millions of possible states and select the move it likes best based upon that. Hmm, the number of chess moves on a board is estimated to be 10^120 a pretty large number. Even Big Blue (Hey another Big!) Had to prune its search trees during the chess matches. Want another example? Computer driving at highways speeds has been successful. You could have a robot car today, except that you couldn't get insurance. (This barrier is now legal instead of technical). This shows a robot can successfully navigate the real world even without having a complete description of it. Have you ever been surprised? That is simply a mismatch between your brain's "picture" of present/future and reality. It startles you, but it doesn't kill you, and it isn't usually a catastrophe. Sometimes your instincts or reflexes take over when your are surprised. Reflexes are relatively easy to build into robots...

You think that's a big number? Have you heard of GO? It has a 19X19 grid to play on. That's 361 spots to place a move, and with all the different variations per move, it beats chess hands down in terms of intuition and strategy.

That's why no computer simulation of GO so far has beaten a real professional GO player (Yes, there are pros for this game, mostly in Asia).

quote:

Originally posted by R Trusedale Hmm, the number of chess moves on a board is estimated to be 10^120 a pretty large number. Even Big Blue (Hey another Big!) Had to prune its search trees during the chess matches. Want another example? Computer driving at highways speeds has been successful. You could have a robot car today, except that you couldn't get insurance. (This barrier is now legal instead of technical). This shows a robot can successfully navigate the real world even without having a complete description of it. Have you ever been surprised? That is simply a mismatch between your brain's "picture" of present/future and reality. It startles you, but it doesn't kill you, and it isn't usually a catastrophe. Sometimes your instincts or reflexes take over when your are surprised. Reflexes are relatively easy to build into robots...

Chess is a finite problem, though. It is a limited number of variables that behave in very specific ways. I don't know how well that would translate to a real-world situation where the possible actions are nearly infinite. That's why I was saying that a robot might be excellent at linear problems (and although chess is complicated, it IS linear) but might not do so well at patterning.

It's hard to say, we have a long way to go. My biggest point has been taken, which is "what assumptions are we making that might not be true?"

quote:

Originally posted by Zola Chess is a finite problem, though. It is a limited number of variables that behave in very specific ways. I don't know how well that would translate to a real-world situation where the possible actions are nearly infinite. That's why I was saying that a robot might be excellent at linear problems (and although chess is complicated, it IS linear) but might not do so well at patterning. It's hard to say, we have a long way to go. My biggest point has been taken, which is "what assumptions are we making that might not be true?"

The human brain itself cannot process near infinite information, and copes with the real world very well. In fact it turns out that the brain takes an amazing number of shortcuts based on expectations. Professional magicians know this, and rely on it for various sleight-of-hand tricks.
The main point I have so far failed to get across is that there is nothing magical about how the human brain processes the world. That means that all the algorithms humans have evolved for coping with the world can and will be translated into silicon, and eventually even surpassed. If you cannot accept the fact that the human brain holds no magic, then there is nothing else I can say that will convince you.
A computer can hold an amazing amount of information about the real world, it need not be naive, and it can learn fast. For an example, please go to http://www.20q.net and try to stump the computer.

Stumped it! Mwahaha.

It depends how closely your answers match the computer's, really.

Hey R Trusedale

Your point didn't miss me. You're right. There *is* nothing magical about our brains . . . they're basically complicated biological computers. And yes, they can easily be fooled by things which do not meet their expectations.

I also agree that robotic brains will eventually become at least the equal of human brains and probably far greater . . . some day. Maybe soon. Maybe not so soon.

The problem I see with that right now is that we humans are only just beginning to understand the specifics of how our brains actually work, and until we do we have little chance to design an equally powerful artificial brain.

quote:

Originally posted by mAc Chaos Stumped it! Mwahaha. It depends how closely your answers match the computer's, really.

The computer's knowledge base is a compendium of millions of human answers to the questions. If you managed to stump it, then its because your answers diverged from what most *humans* know about the topic under question.

Another way to stump it is to come up with a topic it has never encountered. But that is getting pretty difficult.

The third way to stump it is to give nonsensical answers. But this would make a human opponent "give up" too.

Hey, I stumped that thing with "beer" . . . Good fun. But it *is* pretty clever . . . surprised me with a last minute correct guess in another game.

My brother says I should try "viagra" . . . so off I go .

quote:

Originally posted by Black Phoenix Hey R Trusedale Your point didn't miss me. You're right. There *is* nothing magical about our brains . . . they're basically complicated biological computers. And yes, they can easily be fooled by things which do not meet their expectations. I also agree that robotic brains will eventually become at least the equal of human brains and probably far greater . . . some day. Maybe soon. Maybe not so soon. The problem I see with that right now is that we humans are only just beginning to understand the specifics of how our brains actually work, and until we do we have little chance to design an equally powerful artificial brain.

Well let me point out that we still don't know everything about how birds fly. But we were able to design machines that fly far better (farther,faster,higher) than natural birds do. The same might apply to computer thinking. We may eventaully build machines that think differently than humans do, but are our equals in ways that count. In mathematics, most would agree this is true already.
Okay people let me climb down off this soapbox now.....

quote:

Well let me point out that we still don't know everything about how birds fly. But we were able to design machines that fly far better (farther,faster,higher) than natural birds do.

Point conceeded. If we wanted to recreate the human brain, we'd need to know all about it. If we wanted to surpass it, we wouldn't necessarily need to . . .

quote:

Okay people let me climb down off this soapbox now.....

Hey don't worry about it. I'm pretty new here and I already get the feeling that some people think I'm ranting and raving . . . just because I never manage to say what I'm thinking in a remotely concise way . . .

Talking about this stuff with smart people is fun . . . at least as far as I am concerned.

PS

I just tried 20 questions with "anime" as the answer. It actually got it but only after about 30. Still, it's 20th guess was television show. Nifty.

I have flat-out stumped it twice and made it go over 20 questions about 5 times, it won twice. Of course, I am extremely picky about my definitions--one of my items was scotch tape and it asked if it came in a box. It says yes, but I say irrelevant because it may or may not depending on where you buy it (you can often get a plastic package with several rolls of tape in individual holders, no box involved).

Still, it's interesting and does illustrate that simple things can lead to very complex behavior When I spoke of programming a robot, this was the kind of thing I meant, that you would give it some basics and let it "learn" on its own.

This is a great discussion, guys, keep it coming!

quote:

Originally posted by R Trusedale Well let me point out that we still don't know everything about how birds fly. But we were able to design machines that fly far better (farther,faster,higher) than natural birds do. The same might apply to computer thinking. We may eventaully build machines that think differently than humans do, but are our equals in ways that count. In mathematics, most would agree this is true already. Okay people let me climb down off this soapbox now.....

By the same token, we might CHOOSE not to build them quite that smart. It's hard to say right now.