Tick-Tock Man - Next, on Am I Creepy or Not

From: slayerofgod Date: March 2nd, 2009 09:56 pm (UTC) (Link)

Making robots more responsive to human emotion will not stop a Matrix scenario from happening. I say if anything it only increases the likelihood. We're responsive to human emotion, and look how stupid we still are.

(Reply) (Thread)

From: cactuarjoe Date: March 2nd, 2009 10:41 pm (UTC) (Link)

We'd control their emotional responses, though. They wouldn't necissarily be the same as ours.

(Reply) (Parent) (Thread)

From: cal_renfro Date: March 3rd, 2009 12:47 am (UTC) (Link)

This is all so simple. All we have to do is not add that third 'omission of action' law that allows robots to eventually figure out that to save some humans, others must be killed, therefore the 'kill no humans' law does not apply.

And then we can round them up and put them on the robot reservations. Those suckers think we're gonna honor those bogus treaties, but we're totally not!

(Reply) (Parent) (Thread)

From: slayerofgod Date: March 3rd, 2009 01:51 am (UTC) (Link)

Karl nailed it. Humans will never learn from their mistakes.

(Reply) (Parent) (Thread)

From: cactuarjoe Date: March 3rd, 2009 02:22 am (UTC) (Link)

Eh, I don't know about that. But I think "Humans" is an unfair generalization. Humanity as a whole has a habit of repeating the same errors, but societies can and do learn. Not always, but often.

But then this is a bit overly serious of a discussion for a blog entry about Albert Einstein and his battle mech.

(Reply) (Parent) (Thread)

From: slayerofgod Date: March 3rd, 2009 04:18 am (UTC) (Link)

You have any examples of that? I'm failing to think of any major ones.

As far as I'm concerned, our future robot masters are not a joking matter. Humans are too immature a species to be trusted with actually creating a race of beings. The fact that we're creating these robots to be sympathetic to us is only proof of this, and only serves to remind me of something Virgil once said.

(Reply) (Parent) (Thread)

From: cactuarjoe Date: March 3rd, 2009 05:32 am (UTC) (Link)

First, I think it's important to define what I mean by "society learns". What I'm talking about is not the absolute acceptance of an idea or ideal - you simply will never get that in any group of people larger than two. What I'm talking about is a shift in the rules, written or unwritten, that govern a society. Not just laws, but generalized ways of thinking.

By this definition, societies can be anything from a family unit to a major government. Small societies change much faster than large societies, but there have been large-scale shifts in the rules of our society. For example, a hundred years ago, women weren't allowed to vote. The fact that there was a movement in society to change this suggests that there was a definite shift in the ideology and mode of thinking of our society, much in the same way that occurred during the civil rights movement, which continues to this day.

Society, and therefore humanity in general, "learns" by changing in accordance with external and internal stimuli. Or to put it a simpler way, we change with the times. By looking at the past, we can and do learn from our mistakes - not every individual specifically, which is why the same errors often occur, especially in societies led by one powerful individual, such as our own.

Secondly, we're not talking about a new species, we're talking about machines with codifiable perameters. Thinking, largely independent machines, it's true, but we're not talking about subtle, biological entities like Realians here.

As for robot masters, I really think you're expecting too much from a technology that's still decades away from being able to reliably walk up a flight of stairs or smile convincingly, let alone conduct a large-scale war of conquest of its own violition. And there's also Transhumanism to consider. By the point robots can reach our current level of biological sophistication, we'll have long since increased our own abilities through cyberization, computer implants that increase our own awareness and abilities, and possible biological engineering to force evolution in one direction or another, or even abandon our bodies entirely. If anything, we will become robots, not be ruled by them.

But, like I said, it's a bit heavy for such a simple post.

(Reply) (Parent) (Thread)

From: nwashy Date: March 3rd, 2009 12:08 pm (UTC) (Link)

I tend to agree on the point of how close we are to this breakthrough. If anything, the quest toward artificial intelligence has become an exploration of ourselves and an increasing comprehension of just how complex we truly are. I have serious doubts that we'll develop any significantly intelligently autonomous machines that can actually operate generally in the real world until we solve many more of those mysteries.

I think most estimates you might read on our progress in that area are highly optimistic, especially regarding our discoveries just to do something as simple as, say, teach a robot to discern a shadow from an actual obstacle on a video. So far, we've only gotten around that difficult by "training" it through having a human driver drive the machine to give it data on human decisions. Even that ability to learn is highly limited, though, as we had to code in its very ability to learn that specific task. I also imagine there's a big difference between placing a label on an emotion and actually understanding it.

There's a theory I read about in psychology, or perhaps neurology, that believes that emotions are the result of certain areas of the brain becoming more active and certain areas becoming less. If that's true, and it certain seems like a reasonable theory, it does seem we'd have a great deal of control over the emotional responses that would even be available to a robot. In fact, it seems like we'd have to design these emotions deliberately, judging by that theory.

So, would a robot revolution be likely? I doubt it would be due to them reasoning their way out. We would decide what emotion affects in their programming, and it would seem a simple matter to ensure that key directives are unresponsive to that. Could they rewrite them with logic? Not if those core directives are off-limits to any self-modifying code. So unless an artificial intelligence really can rise above its code, that can be defended against.

I think the biggest risks are both human, not AI:

1. Human error in design: We get some priority wrong in the key or ethical directives, or screw up the implementation thereof. A I, Robot scenario is not truly realistic since the first law could not be overridden, provided it is coded correctly. A robot could only reason around that law if we give it the ability to. The problems are more mundane: properly recognizing a human being, properly recognizing dangers to human beings, including those that the robot itself could present. Asimov's laws are in our language, not in the language of the machine. They tell a good story, but the real danger lies in any flaws we may make in the implementation. The robot could only disobey if it can't recognize its own defiance of it.

2. Deliberate design or hacking: This is the one I'd worry most about. If there is going to be an robot war, we'd probably provide the cause by programming it in. This is, I think, the biggest danger of the immaturity you talk about: that someone will deliberately design a robot motivated to destroy or deliberately destroy any safeguards in place (making the SHODAN scenario of System Shock more realistic).

The third possibility is robots somehow gaining a sentient quality we don't yet recognize that would allow them to go beyond their programming. I doubt we'd have any way to predict that without testing the concept. Then again, anyone who believes entirely in naturalism is probably certain that our own intelligence and sentience is derived entirely from the biochemical interactions of our nervous system, suggesting even we cannot overcome our programming but simply have certain ways of altering it. While I do not place all my faith in naturalism and think that we are something more, I have serious doubts about our ability to create that special spark or even the most intelligent robot somehow attaining it. I do acknowledge the possibility because I believe it's arrogant not to, but I think any threat from robots stems entirely from our motivations, not theirs.

(Reply) (Parent) (Thread)

From: animatus Date: March 2nd, 2009 11:20 pm (UTC) (Link)

This battle will be, for lack of a better term, truly epic.

(Reply) (Thread)

From: cactuarjoe Date: March 3rd, 2009 02:25 am (UTC) (Link)

The problem with fighting Neils Bohr is that you can either tell where he is or what he's doing, but not both at once.

(Reply) (Parent) (Thread)