I, for one, welcome our new robot girlfriends.

[Ink Asylum]

I got a good response from my poll about upgrading our bodies with cybernetics, so I figured I'd post a new question.

As artificial intelligence increases in complexity and ability, what combined factors would be enough to convince you that an AI has reached a point where it should be treated as alive and deserving of rights, even if not equal human rights?

For example, we recognize that dogs and other animals aren't as intelligent as human, but they are laws to protect them from cruelty and other abuses. Right now no one would argue that there's anything illegal about killing video game characters, but would there come a point where you would begin to sympathize with a virtual intelligence?

Personally, I know I'd be one of the first to empathize with an AI. I even feel guilty messing with my Sims, and I've never been the kind of person who delighted in torturing them. I still kill loads of characters in video games, but they tend to be pretty simple-minded and don't really exhibit traits that would make my sympathize with them.

[c0m3d14n]

the poll misses the most basic human trait: stupidity ;)

[LarsenNET]

Make me a sandwich and clean the house. That is all.

[Vermillion]

Why does an AI need my empathy? Are we now creating life for the sake of it, or were they brought into existance for a purpose?

In fact, I want a robot that is the exact opposite of human. Emotionless and based upon facts. We have too many stupid, emotionally driven, bags of mostly water running around causing nothing but problems. Why would I want more of that?

For me, it's self awareness and empathy.

[LordDon]

I think self-awareness is enough.

[Frohike]

I think it would be hard for most people to empathize with anything that doesn't have a physical body. The virtual world is still so new that it's still detached from anything resembling life, so it's going to have to housed in some kind of body, robot, whatever.

Hell, people have empathy for trees and plants and all they have going for them is that they are alive and have a physical presenence. Of course there's that whole, "we're fucked if all the plants go" thing too...

[menage]

People get mad and yell at and smash their computer. It's a thin line between love and hate:D

Emotions would do it for me, self awareness is all well and good but without emotions the thing wouldn't care. It wouldn't care about anything at all, not even itself. And in what matter are animals self aware?

[Ink Asylum]

While it might be hard for some people, I think I could empathize with an AI that had no consistent physical body. I'm thinking along the lines of a personal assistant program that jumps between my various devices. One that learns from my behavior so it's faster at predicting what data or program I need at any time. One that I can talk to in a conversational way as a method of user interface.

If a program like that had a believable personality and a natural sounding voice I think I would definitely start to empathize with it, even if it didn't have true self-awareness.

[Ink Asylum]

Emotions would do it for me, self awareness is all well and good but without emotions the thing wouldn't care. It wouldn't care about anything at all, not even itself. And in what matter are animals self aware?

Depends on your definition of "self-aware." Some people feel the mirror test is a good indication. If an animal looks into a mirror and displays an understanding that they're looking at a reflection of themselves and not a different creature that's a form of self-awareness. Recent tests on elephants show that when they look into a mirror and see their reflection with a splotch of white paint on their forehead they will lift their trunk to touch the spot.

[Frohike]

While it might be hard for some people, I think I could empathize with an AI that had no consistent physical body. I'm thinking along the lines of a personal assistant program that jumps between my various devices. One that learns from my behavior so it's faster at predicting what data or program I need at any time. One that I can talk to in a conversational way as a method of user interface.

If a program like that had a believable personality and a natural sounding voice I think I would definitely start to empathize with it, even if it didn't have true self-awareness.

yeah, i would definitely empathize with a program like that. If you can have a natural conversation with it, you can establish a emotional connection even if it's just a program. That's something I'm trying to get into, Computational Linguistics. If you can get a computer to speak fluidly and respond accurately, things with definitely start to change.

[Narradisall]

Id empathise with it more if I knew the moment I accepted the trade agreement it wasn't going to mass armies and send them to clusterfuck my forces.

Especially when the other 3 AI all have the same idea.

[Dukefrukem]

NICE simpsons reference

[Badger]

I think self-awareness is probably the first indicator of sentience, it shows a function we associate with all intelligent beings. On its own it's not enough though, my car is aware that its door is open or that it needs a service but I don't consider it intelligent.

If a being is truly conscious of itself and intelligent enough, its primary motivation should be to stay alive. Therefore self-awareness should go hand-in-hand with self-preservation. This also ties in to emotions and feelings, for example: fear and pain are mechanisms used to aid self-preservation. While I wouldn't consider them necessary functions, they would likely appear as by-products of self-preservation.

So I chose self-preservation, self-awareness and lastly the ability to learn. Taking new experiences and applying that information to problems is the basic recipe for intelligence, and intelligence is the key to building an artificial mind I could empathize with.

I would empathize with a learning, self-aware, self-preserving computer virus as a "living being" long before I would any robotic, rubber coated animatronic puppet trying to emulate human speech.

[Ink Asylum]

When it comes to emotions from artificial intelligence should we be expecting the same emotions that humans display? Many of our emotions derive from the way we've evolved. They allowed us to react properly to threats, to compete for mates, to cooperate and form familial bonds.

The evolutionary track of any potentially sentient AI would be so different from the way flesh and blood animals evolve that it would be unfair to expect the same types of emotions. Fear would probably be the first to show up in any noticeable way.

[TrackZero]

Surprised how few people picked emotions. Without those, how can you have empathy by definition? If the machine doesn't care, why should I?

[TrackZero]

the ability to learn. Taking new experiences and applying that information to problems is the basic recipe for intelligence

Not true at all. If the machine was an AI but lacking the capability to store that memory for any length of time whatsoever, does it stop being an AI? Humans with damage to their ability to make new memories don't stop being human either. Therefore the ability to learn a new technique is bunk with regards to considering it an AI or not.

Though I'd argue that the choice itself is wrong. It needs to be far more granular than the given choices. The capability to learn and the ability to store that knowledge for later use are two completely different things.

[Frohike]

Surprised how few people picked emotions. Without those, how can you have empathy by definition? If the machine doesn't care, why should I?

there's plenty of people out there that empathize with emotionless things. Cars, dolls, Paris Hilton for example

[TheFlyingOrc]

I'll worry about it when it happens. Obviously, current AI doesn't come close to having feelings and personal objectives, which I think would be the bare minimum it would take for it to have rights.

[LongStepMantis]

This isn't an easy one to pin down. I went for self-awareness simply because it would imply that the AI has established an identity for itself. Emotions was my second pick, but they won't have emotion before self-awareness, I wouldn't think. Same with self-preservation. If it wants to continue its "life", it would imply that it has already established an identity for itself, feels emotion, and desires to live.

Imo, self-awareness is the first step in all of these processes.

[Ink Asylum]

I don't know about that. Self-preservation is exhibited by most life, but would you call all life self-aware? A fly has self-preservation instincts, but I don't think it's self-aware.

[LongStepMantis]

I don't know about that. Self-preservation is exhibited by most life, but would you call all life self-aware? A fly has self-preservation instincts, but I don't think it's self-aware.

I see your point. I was thinking of self-preservation in a higher-animal sense. Recognition of a threat, the corresponding emotions (fight, flight, or simply fear) in the desire to prevent death. It didn't help that I keep picturing humanoid robots in my mind, so it makes those instinctual reactions seem more "human". But I agree, self-preservation isn't as key as I initially supposed.

I do still think self-awareness is the key factor, but that's obviously in my opinion.

[Ink Asylum]

It's interesting how few people consider self-replication necessary, considering that, in a purely instinctual sense, it is practically THE driving force of all biological life, including most people.

Perhaps this would be the biggest difference between artificial and biological life. An AI program would have a much higher capacity for personal evolution. In essence, it could debug itself and experiment with new features without having to breed good features into a species or bad ones out.

It's also likely to have a greater lifespan than humans, further reducing a perceived need to reproduce. As long as its accessibly memory is safe from harm and it has access to replacement parts it could live forever. Instead of offspring it would probably be much more interested in maintaining backups.

An AI might view any other intelligent programs as competing for resources. This fear in biological creatures is overriden by reproduction instincts, but coupled with the theoretically infinite life of an AI competition for resources, even amongst its own 'species' or 'family,' might be more important.

[Narradisall]

The only ones I didn't pick were;

a body
self replication
and the AI will never compare.

I would consider all the other essential though

[Nameless]

It's interesting how few people consider self-replication necessary, considering that, in a purely instinctual sense, it is practically THE driving force of all biological life, including most people.

Perhaps this would be the biggest difference between artificial and biological life. An AI program would have a much higher capacity for personal evolution. In essence, it could debug itself and experiment with new features without having to breed good features into a species or bad ones out.

It's also likely to have a greater lifespan than humans, further reducing a perceived need to reproduce. As long as its accessibly memory is safe from harm and it has access to replacement parts it could live forever. Instead of offspring it would probably be much more interested in maintaining backups.

An AI might view any other intelligent programs as competing for resources. This fear in biological creatures is overriden by reproduction instincts, but coupled with the theoretically infinite life of an AI competition for resources, even amongst its own 'species' or 'family,' might be more important.

Exactly what I would have said. Why would self-replication be important if you are for all intents and purposes, immortal. Hardware can break down, but an actual AI could probably live indefinitely, just moving to newer and newer hardware. In fact, it could be dangerous if they could self-replicate...

We do it because death keeps us in check. Without that factor, AIs could outnumber non-AIs in a very short period, depending on their rate of replication.

[tactless]

It wouldn't have to earn my empathy.

[Badger]

Not true at all. If the machine was an AI but lacking the capability to store that memory for any length of time whatsoever, does it stop being an AI? Humans with damage to their ability to make new memories don't stop being human either. Therefore the ability to learn a new technique is bunk with regards to considering it an AI or not.

Though I'd argue that the choice itself is wrong. It needs to be far more granular than the given choices. The capability to learn and the ability to store that knowledge for later use are two completely different things.

I'm confused, how is my statement false? Measuring actual intelligence in machines is fairly subjective, which is why I used examples AI researchers have been using for years: adapting and problem solving.

You're saying that Artificial Intelligence can exist without the ability to learn, and I counter that by saying that isn't showing "intelligence" at all. That is just a database of information. The phrase "Artificial Intelligence" is a misnomer, AI ≠ I.

[TrackZero]

It's interesting how few people consider self-replication necessary, considering that, in a purely instinctual sense, it is practically THE driving force of all biological life, including most people.

I think it's interesting you think that's interesting. Why would it be a factor at all? If someone can't have kids for any reason, they don't stop being human. From the original question, I don't see why your empathy for a AI would disappear because it can't create new AIs. It has no bearing.

Perhaps this would be the biggest difference between artificial and biological life. An AI program would have a much higher capacity for personal evolution. In essence, it could debug itself and experiment with new features without having to breed good features into a species or bad ones out.

It's also likely to have a greater lifespan than humans, further reducing a perceived need to reproduce. As long as its accessibly memory is safe from harm and it has access to replacement parts it could live forever. Instead of offspring it would probably be much more interested in maintaining backups.

Both of those points are assumptions (self-modification of code and lifespan) that sci-fi writers have more deeply explored in recent years. Which don't always hold true based on certain conditions.

With regards to self-modification, if you're assuming this AI has at least human level intelligence it'd have to ask itself whether it wants to become another "person" entirely. You're not considering the big picture that self-modification (especially if the code was open, readable and recognizable) isn't a cut and dry decision. You're thinking of it in a biological context. i.e. What if I had green eyes instead? That'd be great, you could do so without much concern. Now think about brain modification. In a human that's a big deal. You're talking about changing the very way you think, the person you are, losing yourself. Even if it just came to making you brain run faster, it changes the very perception of the world and you cease being what you were before. Now if you look at it from a pure AI perspective, we're really only talking about software. That's like being able to reach into your mind, change your very thinking patterns and memories. Who says the AI would want to do that? IMHO, that'd be a dangerous, scary and slippery slope and may go against (if it has it) it's own self-preservation.

Also lifespan. There's so many variables to consider there. Is the AI only staying on the one piece of hardware it has? Can it transfer to another computer (and before you say "of course", slow down cowboy and think it through)? How protected is it's system? Memory corruption with current technology happens constantly in nature, even the most stable systems will eventually have enough bits flip the wrong way to become unstable and crash. That's not even taking into account the sheer size the AI's core system may have to be (not accounting for it's "memory"). Most operating systems (not even counting whatever applications are running on them) do not stay up that long, with the world record being 11 years (and one one rumored at 18). That's a short lifespan compared to us. That's also not considering bugs in it's own code that could cause self-corruption before it managed to correct them (if it even would choose to modify itself, again) or it isn't aware of.

An AI might view any other intelligent programs as competing for resources. This fear in biological creatures is overridden by reproduction instincts, but coupled with the theoretically infinite life of an AI competition for resources, even amongst its own 'species' or 'family,' might be more important.

That's possible, though that's trying to predict behaviour on a non-existent non-biological species that could be developed any number of ways.

Honestly, your view of AI theory seems (and I'm not trying to be offensive here) very 1980s/early 90s. I can point you to some newer works that have explored some of the things I've pointed out so far.

[Vermillion]

There is a difference between empathy and sympathy.

Empaty is an understanding of feelings, sympathy is an emotional response to it. I want it to simply empathize with my emotions (take them into consideration) and do the logically correct thing.

And I don't think they need a physical body. Just look at Cortona. I mean, yowaza. You can corporial my being any time.

Feeling robots is the stupid. All I need to hear when I come home is my wife AND my 360 complaining that I don't spend enough time with them. The PS3 already knows that it's adopted and will never truly be loved.

[ShivaX]

Exactly what I would have said. Why would self-replication be important if you are for all intents and purposes, immortal. Hardware can break down, but an actual AI could probably live indefinitely, just moving to newer and newer hardware. In fact, it could be dangerous if they could self-replicate...

We do it because death keeps us in check. Without that factor, AIs could outnumber non-AIs in a very short period, depending on their rate of replication.

Well I can't see how self-replication would even factor in with AIs. I mean the idea of self-replicating nanotechnology is out there. That doesn't require any sort of intelligence, just a routine that says something along the lines of "If you have X materials create a copy of yourself."

Most computer viruses are self-replicating and I'm not about to decide they're worth saving from destruction because of it.

[TrackZero]

I'm confused, how is my statement false? Measuring actual intelligence in machines is fairly subjective, which is why I used examples AI researchers have been using for years: adapting and problem solving.

You're saying that Artificial Intelligence can exist without the ability to learn, and I counter that by saying that isn't showing "intelligence" at all. That is just a database of information. The phrase "Artificial Intelligence" is a misnomer, AI ≠ I.

I'm saying that learning is a split process. The actual ability to discover and new technique and the ability to use that new technique at a later time. If that technique cannot be stored either due to hardware or software constrictions, it does not mean that the AI isn't intelligent just because it forgot how to do it. Nor would anyone's empathy be affected by the inability to do so (in fact, pity would likely occur).

Your statement "Taking new experiences and applying that information to problems is the basic recipe for intelligence" is false. Not your intent, which is to say that subjective reasoning is a cornerstone for intelligence, that's true. But saying that new experiences need to be re-applied at a later time for problem solving is false.

If you want to debate semantics of the term AI, I'll leave you to that.

[Wasson_]

to gain my "empathy" huh? for an AI to have my empathy it would first need the inability to control it's emotions completely and then the ability to make mistakes often enough for it to matter.

[Nameless]

Well I can't see how self-replication would even factor in with AIs. I mean the idea of self-replicating nanotechnology is out there. That doesn't require any sort of intelligence, just a routine that says something along the lines of "If you have X materials create a copy of yourself."

Most computer viruses are self-replicating and I'm not about to decide they're worth saving from destruction because of it.

And this is why AIs are scary. I'm sure they'll destroy us if they ever are created. I have difficulty feeling empathy for something that will almost undoubtedly end our race's existence.

It's cool if they're like Data, but less so if they're like the Borg.

[Troggles]

It needs to give me blowjobs before I give it any empathy.

[Ink Asylum]

I think it's interesting you think that's interesting. Why would it be a factor at all? If someone can't have kids for any reason, they don't stop being human. From the original question, I don't see why your empathy for a AI would disappear because it can't create new AIs. It has no bearing.

It's interesting because it would mean empathy with a new form of life, not even just a new biological species, that has motivations far different than those that drive pretty much every other life form. Humans who can't or don't have kids are still more similar to their other people than they are different.

One of the ways that we empathize, or sympathize which is perhaps a more accurate term for my original question, is by observing how similar their behavior is to ours. We relate more to animals that have family units and raise their young with care and affection.

Being able to relate to a form of life so alien that it has no need to reproduce, a desire that influences so many of our species' behaviors, requires a much more open mind than I would give the average human credit for. So what I find interesting would be how mature we all are. :)

With regards to self-modification, if you're assuming this AI has at least human level intelligence it'd have to ask itself whether it wants to become another "person" entirely. You're not considering the big picture that self-modification (especially if the code was open, readable and recognizable) isn't a cut and dry decision. You're thinking of it in a biological context. i.e. What if I had green eyes instead? That'd be great, you could do so without much concern. Now think about brain modification. In a human that's a big deal. You're talking about changing the very way you think, the person you are, losing yourself. Even if it just came to making you brain run faster, it changes the very perception of the world and you cease being what you were before. Now if you look at it from a pure AI perspective, we're really only talking about software. That's like being able to reach into your mind, change your very thinking patterns and memories. Who says the AI would want to do that? IMHO, that'd be a dangerous, scary and slippery slope and may go against (if it has it) it's own self-preservation.

Making such drastic changes would be difficult for any intelligent creature to do, but the potential is there, in a greater way than it is with humans or any other biological life. Where there is potential for self-improvement it's very likely someone will try, even if other people see it as dangerous. People regularly take performance enhancing drugs that other people see as self-destructive. My prior thread on cybernetic body modifications shows that a lot of people would make drastic changes to their body given the opportunity. An AI may or may not be willing to make such changes, but that's a lot better position to be in than being mostly incapable of doing so.

Besides, even without making modifications to its 'brain,' whatever you would consider the core of an AI's being or personality, it still would be much more capable of making changes to its 'body' that would allow it to evolve and increase its abilities than any life before now. Like it or not I'm stuck with this fragile human body that has a top speed, weight capacity, lung capacity, etc all with upper limits of how I can improve them. I don't think you would say that if I could hold my breath for a half hour, which would mostly require modifications to my lungs and circulatory system, that it would mean such a great change in what my personality and sense of self would be.

Also lifespan.

Which is why I said a program would probably be more inclined to create backups and failsafes rather than be interested in creating completely new programs that would continue on after it died. Humans don't really have that option, which is why reproducing is often thought of as a way for a mortal person to continue their legacy.

Honestly, your view of AI theory seems (and I'm not trying to be offensive here) very 1980s/early 90s. I can point you to some newer works that have explored some of the things I've pointed out so far.

None taken. Feel free to post links to theories or fiction that explores these ideas in more detail. Always open to new reading.

[torrefaction]

Wait...only four of you think a robot girlfriend would just not be right?

FOUR?

[Ink Asylum]

And this is why AIs are scary. I'm sure they'll destroy us if they ever are created. I have difficulty feeling empathy for something that will almost undoubtedly end our race's existence.

Welcome to the universe of Mass Effect, where AI research is banned for just that reason. It led to the Geth, which are pretty much their version of the Borg.

[ShivaX]

Wait...only four of you think a robot girlfriend would just not be right?

FOUR?

I think you missed the point. :p

[ShivaX]

Welcome to the universe of Mass Effect, where AI research is banned for just that reason. It led to the Geth, which are pretty much their version of the Borg.

Same goes for the Warhammer 40k universe iirc.

[Primus]

Wait...only four of you think a robot girlfriend would just not be right?

FOUR?

Robot girlfriend would have all the pluses with none of the negatives of a real woman.

[Ink Asylum]

For once, "Sucks like a Hoover," wouldn't be a metaphor.

[Badger]

I'm saying that learning is a split process. The actual ability to discover and new technique and the ability to use that new technique at a later time. If that technique cannot be stored either due to hardware or software constrictions, it does not mean that the AI isn't intelligent just because it forgot how to do it. Nor would anyone's empathy be affected by the inability to do so (in fact, pity would likely occur).

Your statement "Taking new experiences and applying that information to problems is the basic recipe for intelligence" is false. Not your intent, which is to say that subjective reasoning is a cornerstone for intelligence, that's true. But saying that new experiences need to be re-applied at a later time for problem solving is false.

If you want to debate semantics of the term AI, I'll leave you to that.

Ok, I see what you're saying. I agree my wording is rubbish, glad you saw the idea I was trying to put across.

So say there are many different functions involved in learning, and if an artificial mind is incapable of any one of them it doesn't make them any less intelligent. This I completely agree with. Now if it was incapable of every function of learning, that is when I would cease to consider it intelligent.

I don't draw a parallel with a human incapable of every function of learning, as a biological entity with intuition, emotion, feelings, personality, ancestory etc. I'm already wired for empathy. With a machine, that's not the case.

[boratika]

the poll misses the most basic human trait: stupidity ;)

I think you mean Schadenfreude.


Of course, one AIs have that, we are doomed.

[TrackZero]

Besides, even without making modifications to its 'brain,' whatever you would consider the core of an AI's being or personality, it still would be much more capable of making changes to its 'body' that would allow it to evolve and increase its abilities than any life before now. Like it or not I'm stuck with this fragile human body that has a top speed, weight capacity, lung capacity, etc all with upper limits of how I can improve them. I don't think you would say that if I could hold my breath for a half hour, which would mostly require modifications to my lungs and circulatory system, that it would mean such a great change in what my personality and sense of self would be.

Again, for an AI there is no body, if there was, it's irrelevant to the question. The only thing the AI has is it's own software, which was my original point on how it would even attempt to modify itself if it wanted to.

None taken. Feel free to post links to theories or fiction that explores these ideas in more detail. Always open to new reading.

Check out:

Capacity
Divergenge
Halo: First Strike (no, I'm not kidding)
Starfish
Maelstrom
Behemoth (B-Max and Seppuku)
River of Gods

[TrackZero]

So say there are many different functions involved in learning, and if an artificial mind is incapable of any one of them it doesn't make them any less intelligent. This I completely agree with. Now if it was incapable of every function of learning, that is when I would cease to consider it intelligent.


Good stuff, I agree with that. If it can't make reasonable observations and conclusions to understand new ideas, I wouldn't consider it intelligent, at least not on anything other than an insect level.