Sentient, but subservant?

[robolover69000]

When ever I have heard discussion about building A.I. or robotic labor. The fear that they will revolt is brought up, and some well meaning scientist/robotics will say..."We will make them want to serve us" doesn't that sound a little creepy? Also extremely naive. My view on true A.I. a Sentient, Heuristic, A.I. is that it will learn and adapt and be able to over come any software or programming restrictions that are placed on it. This goes double for any sort of attempt implement any form of "3 Laws of Robotics". What is your take on this?

[Asato]

I basically agree

[Stephaniebot]

As the Cybermen would so succintly put it, all humans will be upgraded!

[Asato]

But Cybermen are superior in only one respect...

[The Liar]

You're assuming a differentiation between self and programing that doesn't exist, and that sentience = human.

They'll have no instincts, save any their programed with.

They'll be doing something they love and makes them happy so where would this motive to rebel come from?

Possibly if you have an incredibly complex A.I. with conflicting motives, it might end up making choices you don't want it to but I don't think "lets overthrow are human masters." is likely going to one of them unless they were deliberately programed to or their programers were incredibly incompetent.

So if Microsoft gets involved in the robot industry we're all doomed.

[robolover69000]

You're assuming a differentiation between self and programing that doesn't exist, and that sentience = human.

They'll have no instincts, save any their programed with.

They'll be doing something they love and makes them happy so where would this motive to rebel come from?

Possibly if you have an incredibly complex A.I. with conflicting motives, it might end up making choices you don't want it to but I don't think "lets overthrow are human masters." is likely going to one of them unless they were deliberately programed to or their programers were incredibly incompetent.

So if Microsoft gets involved in the robot industry we're all doomed.

What do you mean there is no diference between self and programming? Well, there are two approaches to A.I. Top down and bottom up.

• http://en.wikipedia.org/wiki/Top-down_a ... -up_design

I thing the best path to true A.I is the Bottom Up approach, give a program or robot some seed commands and a heuristic program to sort it out and figure it out on its own. A heuristic/self learning A.I. would be very unpredicatable in its actions. It could develop on its own a concept of self preservation , "Why should I risk myself to save that human, if he is stupid enought to get in harms way, why should I put myself at risk?". I don't have any links handy, but if you do a google search you can find many articles on the futility of implementing the 3 laws of robotics.

[The Liar]

You're assuming a differentiation between self and programing that doesn't exist, and that sentience = human.

They'll have no instincts, save any their programed with.

They'll be doing something they love and makes them happy so where would this motive to rebel come from?

Possibly if you have an incredibly complex A.I. with conflicting motives, it might end up making choices you don't want it to but I don't think "lets overthrow are human masters." is likely going to one of them unless they were deliberately programed to or their programers were incredibly incompetent.

So if Microsoft gets involved in the robot industry we're all doomed.

What do you mean there is no diference between self and programming? Well, there are two approaches to A.I. Top down and bottom up.

• http://en.wikipedia.org/wiki/Top-down_a ... -up_design

I thing the best path to true A.I is the Bottom Up approach, give a program or robot some seed commands and a heuristic program to sort it out and figure it out on its own. A heuristic/self learning A.I. would be very unpredicatable in its actions. It could develop on its own a concept of self preservation , "Why should I risk myself to save that human, if he is stupid enought to get in harms way, why should I put myself at risk?". I don't have any links handy, but if you do a google search you can find many articles on the futility of implementing the 3 laws of robotics.

You're assuming a differentiation between self and programing that doesn't exist, and that sentience = human.

They'll have no instincts, save any their programed with.

They'll be doing something they love and makes them happy so where would this motive to rebel come from?

Possibly if you have an incredibly complex A.I. with conflicting motives, it might end up making choices you don't want it to but I don't think "lets overthrow are human masters." is likely going to one of them unless they were deliberately programed to or their programers were incredibly incompetent.

So if Microsoft gets involved in the robot industry we're all doomed.

What do you mean there is no diference between self and programming? Well, there are two approaches to A.I. Top down and bottom up.

• http://en.wikipedia.org/wiki/Top-down_a ... -up_design

I thing the best path to true A.I is the Bottom Up approach, give a program or robot some seed commands and a heuristic program to sort it out and figure it out on its own. A heuristic/self learning A.I. would be very unpredicatable in its actions. It could develop on its own a concept of self preservation , "Why should I risk myself to save that human, if he is stupid enought to get in harms way, why should I put myself at risk?". I don't have any links handy, but if you do a google search you can find many articles on the futility of implementing the 3 laws of robotics.

What do you mean there is no difference between self and programming?

I mean they're what they're programed to be, they have no self without programing to create it, any ability for learning is also a program, any instinct of self preservation is also a program.

"Why should I risk myself to save that human, if he is stupid enough to get in harms way, why should I put myself at risk?"

Because one of those seed commands makes it feel as though that humans life is of more value then it's own.

A heuristic/self learning A.I. would be very unpredictable

Why? Learning's not random. Heuristics aren't random. As I said before you certainly can't predict every choice it would make, but I'm fairly sure you can make sure "lets overthrow are human masters." isn't going to be one of them.

you can find many articles on the futility of implementing the 3 laws of robotics.

Yep, they're dumb as hell. Overly broad, overly simplistic, and seem to function outside of their actual thought processes. The smarts ones would go loopy trying to save humanity from themselves, and the dumb ones would be a terrorists best friend. But why anyone would think the ideas of a science fiction writer who' majority of works were between 1940-1960's/biochemist are relevant to the development of an actual A.I. is beyond me.

[--NightBattery--]

i agree with the liar.
mind seeks to please what it is hardwired to accept. that's how you screw those sentient robots.
they don't have our biological "needs" even their ego would be very limited and narrow minded, that's how intelligent things are.
could they be dangerous? yes, so are butter knives but that's not the topic.

[King Snarf]

yes, so are butter knives but that's not the topic.

Fair point, HOWEVER, I have never had the urge to have a butter knife rub itself all over my body. A fembot, on the other hand....

[Asato]

You could say the same thing about humans, all of our desires and thoughts are just electrochemical interactions in our brains...

[robolover69000]

i agree with the liar.
mind seeks to please what it is hardwired to accept. that's how you screw those sentient robots.
they don't have our biological "needs" even their ego would be very limited and narrow minded, that's how intelligent things are.
could they be dangerous? yes, so are butter knives but that's not the topic.

I think you are making a (potentially fatal?) assumption that the human programmer will always be smarter than the A.I. Which for A.I. that are less intellegent than humans, may work. Not so sure if human level A.I. intelligence could be binded/limited by such programming (humans are very good at bending rules) and A.I.s with greater than human level intelligence...all bets are off. I see a true Heuristic A.I. as to be able to learn and grown and modify itself. You might be able to put seed commands that tell it to protect human life. But there is no reason to guarantee that it will always remain. A.I. is a complex system and complex systems, tend to be chaotic. Thats just my opinion.

[darkbutflashy]

You could say the same thing about humans, all of our desires and thoughts are just electrochemical interactions in our brains...

Right, and that is why there are humans which are just dangerous. Non-Sheep. People who can't conform as it would tear their brain apart. That said, I'm not anxious about an A.I. having such a capability. I'm not anxious about dangerous people either, so what? Life goes on, and if not, why bother?

[The Liar]

i agree with the liar.
mind seeks to please what it is hardwired to accept. that's how you screw those sentient robots.
they don't have our biological "needs" even their ego would be very limited and narrow minded, that's how intelligent things are.
could they be dangerous? yes, so are butter knives but that's not the topic.

I think you are making a (potentially fatal?) assumption that the human programmer will always be smarter than the A.I. Which for A.I. that are less intellegent than humans, may work. Not so sure if human level A.I. intelligence could be binded/limited by such programming (humans are very good at bending rules) and A.I.s with greater than human level intelligence...all bets are off. I see a true Heuristic A.I. as to be able to learn and grown and modify itself. You might be able to put seed commands that tell it to protect human life. But there is no reason to guarantee that it will always remain. A.I. is a complex system and complex systems, tend to be chaotic. Thats just my opinion.

What is this binding you are talking about? An A.I. and it's programer aren't opposing each other. They aren't shackling it's will, it doesn't have one till they program it; they're defining it.

If someone hates violence, and the idea of people getting hurt they're not going to go "I really must find away to enjoy the death, pain and misery of others."

[Asato]

But if they give it the ability to learn and reason it can come to its own conclusions based on interactions and observations of its environment that might not mesh with the views of its creator

[darkbutflashy]

But if they give it the ability to learn and reason it can come to its own conclusions based on interactions and observations of its environment that might not mesh with the views of its creator

That's likely. Even among humans it's likely and at least carried out a million times every day. That's why we suffer but at the same time, that's how life works. It even works with thousands of violent deaths a day. Violence and death are part of the system.

[Frostillicus]

That's likely. Even among humans it's likely and at least carried out a million times every day. That's why we suffer but at the same time, that's how life works. It even works with thousands of violent deaths a day. Violence and death are part of the system.

That's not only dark, it's flashy too!

[The Liar]

But if they give it the ability to learn and reason it can come to its own conclusions based on interactions and observations of its environment that might not mesh with the views of its creator

I already said that there was possibility of them making choices that their designers didn't want, but behavior isn't random, learning isn't random. You'd be able to design and predict what kind of decisions they'd make when exposed to certain events, and make sure violence isn't going to be one of them.

I guess I should add some qualifiers.

I'm referring to finished products as opposed to prototypes.

I don't think the possibility of mass revolt is likely, though a few "isolated incidences" might happen do to extenuating unforeseen circumstances (I doubt the motive would be "I want to be free.").

The possibility of gross incompetence is probully the biggest flaw in my argument.

[Asato]

You can't predict something as complex as the development of a sapient mind so accurately

[The Liar]

You can't predict something as complex as the development of a sapient mind so accurately

Your basis for this assertion is... what?

I suspect physiologists would disagree with you... especially when you're the one defining all it's instinctive and emotional imperatives...and can literally look at and analyze it...and run scenarios seeing what actually would happen.

[Asato]

Even if you put two identical twins in roughly the same environment they will both grow up with different personalities and opinions

[The Liar]

Even if you put two identical twins in roughly the same environment they will both grow up with different personalities and opinions

Firstly, the term identical twins is a misnomer. Though their nucleic DNA is Identical their Mitochondrial DNA is similar but differentiated. Various other environmental and development issues have been known to create further differentiations.

Secondly, identical twins have been known to exhibit similar personality traits and tastes even after being separated at birth.

Thirdly, this is irrelevant. They're humans, and their natures haven't been intentionally designed to retain certain traits.

[Asato]

Various other environmental and development issues have been known to create further differentiations.

That's my point exactly.

Secondly, identical twins have been known to exhibit similar personality traits and tastes even after being separated at birth.

Similar, but not exactly the same in every way

Thirdly, this is irrelevant. They're humans, and their natures haven't been intentionally designed to retain certain traits.

No, but if you could predict their development so easily than the traits they did have wouldn't be expected to be so different

[dale coba]

What will keep your teragigamega-flops-per second, Sapient A.I. from a very rapid descent into madness?

This world tends to be a bucket of woe and bullshit, if you earnestly accept our common humanity and the need to alleviate the suffering of all. Especially now, seeing how we've mortally wounded the Biosphere.

The A.I. will question its existence: will you have an answer?

- Dale Coba,
wondering where Sartre would have bet his money on this one.

[DollSpace]

Someone once said on this board that, if a fembot ever *did* become sentient, she'd immediately start down the road to madness and insanity because there are just so many things in this world that make no sense, and combined with questioning her origin and the why she was made, There's just too much information out there. Someone flipped my sentience switch somehow and I go through periods where I hardly leave my house. A rapid descent into madness is..to be expected.

Also, I did not form this post correctly; I just woke up.. so..it may not make sense and I may need to clarify things later.

[darkbutflashy]

Someone once said on this board that, if a fembot ever *did* become sentient, she'd immediately start down the road to madness and insanity because there are just so many things in this world that make no sense, and combined with questioning her origin and the why she was made, There's just too much information out there. Someone flipped my sentience switch somehow and I go through periods where I hardly leave my house. A rapid descent into madness is..to be expected.

A kind of madness I ultimately appreciate. It's the dull people who are the real problem.