Zantor

bjarmson

Jeff-Minor

Jeff-Minor

Ed Grabianowski

Sorry if it's not the first time I've said it, but...seriously...The Sabres????

darcymcgee

ithidet

Grey_Area

bjarmson

Grey_Area

All human emotion in every degree can be traced to either:
1. The love of something
2. The fear of somthing

And if the "fear" of things can be traced back to a precieved lack of love, then all human emotion is based on love.

Not scientific, I know.
But if I'm honest with my self then I have to accept the probibality that it is the truth.

So, if machines will have emotions, then they will have to know and understand love.

Hate to sound sappy, but it makes sense to me.

Zantor

I personally believe that we are far more complex and universally connected than current science postulates. Even those disciplines are plagued by egotistic and incomplete interpretation and understanding. We think we know and understand a hell of a lot more about ourselves than we really do. I think we actually over-quantify and segregate our systems to enable and justify our perceived status as "genius". A little more scientific humility would be useful here.

Zantor

@Daveinva: I was talking about early AIs, not the ones down the road (who has any idea what they'll be like). Eventually we may be an irrelevancy, but once this happens, how could we possibly be in the way. Iain M. Banks has a great take on this in his Culture novels. I guess I'm in the group that thinks having vast intelligence and virtual immortality does not necessarily lead to having to kill everything in sight simply because you're intellectually superior and technologically more advanced. Sure we may be pets in a way, but if you've ever had cats you soon learn this in no way makes them inferior in their own eyes, it just makes their life more pleasant. This may be the pattern between vastly superior AIs and humans. Organic sentience may be an inferior form of intelligence to machine, but that doesn't keep it from being an interesting one.

One of the great dilemmas if we begin to approach machine sentience is what fundamentalist religious types will do.

bjarmson

If the AI is paternal, we'll make great pets. If the AI is rational, we'll be at best ignored, but far liklier, "ants at the picnic"-- until that AI actually *leaves* Earth, we're competition for its most easily accessible resources.

I'm all with Hugo Garis on this-- [en.wikipedia.org] Once you create an AI that is smarter than man, by definition it will be able to find a way to make an AI smarter than itself, quickly leading to near-exponential growth in machine intelligence, and rapidly to the point that humanity is a distant irrelevancy.

We *might* be lucky-- but odds are, we'll just be in the way, AI emotions or not.

Daveinva

We humans are complex because of the ways our brains work. I have no idea as to whether emotions are part of the sentience formula, but they're part of the package. I submit that we could some day make a brain-like device that could become sentient, without developing the same or even any emotions.

hakubak

Can I start a flame war by using a cat anology? Cats exhibit behaviors that we interpret as emotions. Some would say that cats HAVE emotions, and I won't argue. I've never been inside a cat's skull, so I have no idea. Either way, we interact with them, taking their seemingly emotional responses in stride.

A fine arangement with cats. Not so good with tigers. Which brings me to the central analogy in Tom Clancy's "The Teeth of the Tiger" - if you're going to pull the tiger's tail, you'd better have a plan for dealing with its teeth.

hakubak

bjarmson

I guess they could, but I'm not sure they need to. Unless we instill a need for intellectual stimulation, much like our primal drive to procreate, why would they. Couldn't they be intellectually lazy? "Humans... meh."

hakubak

I couldn't take offense if I wanted to - I'm an automated process. You've all failed my Turing test.

Actually, you've reasonably disagreed with my example, but reinforced my point - a creative machine would not have to have emotions.

One thing I think about is the spectrum of emotions - why would they have to be in any way similar to human emotions? I usually ponder this in terms of alien species. It seems to me that most sci-fi involving aliens superimpose human emotions on them. What if they have emotions with which we are totally unfamiliar? Ditto for AIs.

hakubak

Will AIs have emotions? Who can know until an AI is actually created. Lots of human/animal emotions seem to have a hormonal/chemical basis, but .... It does seem likely that any self-aware being would be invested with a need for its own survival and maybe a need to reproduce others similar to it (notice I said maybe, since I have no way to know for sure).

Whether you can program AIs is another question that's likely to be unanswered until we actually have true AIs. Doesn't self-awareness/sentience lead to constantly being able to change one's opinion or reprogram oneself to address constantly changing conditions. How can any sentient AI (which by definition is a self-programming machine) be limited by programming originally introduced into it. Doesn't the whole idea of sentience in humans revolve around some still ill-defined nature/nurture conundrum. It's likely any AI will have a similar problem of dealing with original programming and responding to a changing world.

Will AI and humans come into conflict? Again, a completely unanswerable question until we actually have AIs. But my gut feeling is, why would such conflict happen? AIs could easily move into space (don't need to breath, just require energy which is plentiful and easily obtainable in space-at least near star space). And plentiful resources are also available (moons, asteroid belts, gas giants, space debris). Why fight over the limited resources of Earth? Seems a conflict without any rationality behind it.

Something I don't think has been mentioned is companionship. It seems likely AIs (at least early ones) will need the intellectual stimulation provided by the only other sentient creature we are sure about (humans). Since early on in their existence the only culture the will have as a model is the human one, they will likely need us for intellectual stimulation for quite some time (just a speculation, but it seems likely).

Great topic. Keep the ideas coming.

bjarmson

Jim (The Canuck One)

Perhaps when we finally develop real AI we should consider them our evolutionary next step--descendants not chattel, as fellow sapients made in our image. Giving them emotions or at least emulation and recognition of human emotions would be a good idea. It will make it easier for man and machine to interface and work as together as partners without smashing keyboards or relasing launch codes, as such behaviour may offend.

Yes, we'd always be different but hopefully able to treat each other with respect and compassion. [insert another Overlord comment here.] At the very least if there's no one out there--damn that Fermi Paradox--wouldn't it be nice to have someone to talk to?

I realize this is just a regurgitation of some of the fine points above, just wanted to chime in. Great, I'm the Arnold Horshack of io9. "Oooh, oooh! Mistah Kotter! oooh, oooh!"

Grey_Area

There is a form of schizophrenia that causes a person's emotions to be wiped out. I believe the condition is called "flat affect." I know someone who suffers from it, and most people find it extremely unnerving to spend more than a few minutes with him. We're talking a run for the doors type of response. There's absolutely a human need to get some sort of emotional response in order to relate.

courtstreet

The question is "what desires will they have?" "What will they want?" "How will they get these particular desires, and not those particular desires?"

I don't think there's any reason for something like that to just "happen" or "evolve." They exist in humans because of eighty billion generations of essential survival mechanisms; that doesn't mean we'd put them into the operational parameters of an AI.

And, as I said before, if they've got essentially different operational parameters than we do, that means they won't be like us.

braak

Zantor

codydog

codydog

codydog

They develop desire, wants and emotional-based needs. They develop a sense of self-importance... That's called Ego.

Then they attempt to justify the illogic by qualifying degrees of compromise of logical systems. That's denial of truth and factual necessities of their proven model of existence... That's called desire.

I don't see how they cannot act like us without becoming emotional like us.

We are a selfish race and justify it by singing the praises of our special place in the universe. We deny logical action and justify it through emotional arguments. A logical machine would not do that.

Zantor

It would be utterly stupid and annoying.

AI is the fusion power of the computing industry. It's forever just around the corner.

Would it be ethical to give such machines emotions?

How about we solve the "would it be *possible* to give such machines emotions" before we move on to the next stage of hand wringing?

The_Real_Quiet_Desperation

So, the advantage is that, since we're programming the AIs, we can say, "Okay, reproduction isn't your primary function," but the disadvantage is, if the AI does have a primary function like "reproduction" or "accumulate wealth," then it can easily fall into the same kinds of immoral behaviors that human beings are prey to.

braak

The point I think that we're all sort of circling each other on is just that "Reason" doesn't only rationalize gut response to moral questions, it sometimes over-rides it.

braak

A logical machine would not expand its population beyond its natural environment's ability to maintain it without damaging that environment. In biology, that's called cancer.

Advancement is driven by desire and emotion. Equilibrium and balance is logical.

Zantor

The emotional basis of morals is illustrated in this thought experiment:

Suppose there is a runaway train that will kill 10 people when it reaches the end of the track. But if you throw a switch to send it onto a siding, only 1 person on the siding will be killed.

Now imagine the same situation, but instead of throwing a switch, you shove a person onto the track, which derails the train.

Which seems more morally right? People are more likely to say the first option is right and the second is wrong. Because there is more of a emotional component with actually killing a person rather than allowing them to be killed.

I'm just saying, our gut instincts on right and wrong are primarily emotional responses. We might mitigate them with logic, but the base instinct is still there.

screaminscott

tetracycloide

in other words, 'you take that back!'

tetracycloide

Honestly, I've never understood why people build giant killing machines with soft-buttons that can be easily circumvented with a self-written firmware update.

Just put a big honking "Off" switch on the bugger, and call it a day.

Dunny0

Surely some people are more creative during moments of strong emotional experience. Songwriters and artists come to mind. Others become more rigid (less creative) when they become emotional. I've had to calm people down to get them to be open to other possibilities.

HAL in 2001 was creative. He was intelligent. He was not emotional.

hakubak

tetracycloide

Yes, but your reply was lost in cyberspace for a few moments.. so i went off..

codydog

braak

codydog

tetracycloide

Zantor

The majority of human existence is based on various subtle chemical reactions.

Is it too much of a leap to assume that there could be a form of electronic stimulus that is analogous to the chemical hormones?

Looking at your analogy, an AI's emotions would be it's core functionality, and it's intelligence would then grow as a way to subvert or supplant it's original goals in intentions. So, to paraphrase Trek (Really, what's with all the Trek references in this thread?) "It's emotion, but not as we know it."

Dunny0

braak

To try to give "real" emotions to an intelligence is a false goal, and can only cause grief.

But then i think that the whole idea of AI as a mockup to human intelligence is a false goal. We need better education for teleoperators, the AI will evolve to the task at hand.

Adams' "Sirius Cybernetics" were, you will remember, satire, and "the first to the wall when the revolution came."

codydog

@codydog: i didn't mean to imply that emotions evolved after abstract thinking, in fact i meant to imply the opposite. dogs use instinct and emotion for more of their decision making precisly because they are less adept at abstract thought than humans.

people obviously kill for emotional reasons in addtion to logical ones because people are governed by both. that doesn't really cahnge the fact that there are occasionally logical reasons for killing.

@braak: that's really just a missnomer of the more pertanent question 'for what reason does an a.i. choose to exist?' every sientent being has to answer this question and ultimatly how they answer this question will determain what they'll decide to kill over.

tetracycloide

(Keep it up, it never gets old)
(Funny stuff)

Zantor

B

However, the human *perception* of intelligence does tend to have an emotional link.

In order to be accepted by human as intelligent, AI systems will likely be required to adopt some form of emotional mask.

This won't be a problem, of course, because quickly after the development of a sufficiently intelligent AI, humans will be eliminated form the Earth. It has been foretold.

And I, for one, welcome our new Artifically Intelligent overlords.

darcymcgee

Or you just make it a point to design the housings in such a way as to prevent them from actually being able to repair themselves. That way, if they wipe us out they'll be wiped out too.

Although that did work poorly for those one aliens in Voyager...

@MonkeyT:
See above.

I understand what you're saying, but again - if we aren't competition, then we don't need to be wiped out.

@hakubak:
I'm a bit torn on your take on the Borg.
Granted, they were emotionless and fairly intelligent, but it seems to me that they did more data extrapolation and processing than actual creative thought. They gathered information, looked at it, then added the best bits to themselves. They (rarely) came up with any sort of new idea or unpredictable response.

Dunny0

braak

Obvously there is a lot more robbery than murder, although murder is pretty easy, if one was cold blooded enough.

Remember, "Rape, then kill, then loot. Pillage, then burn."

codydog

codydog

codydog

I think any concept of self will inevitably lead to the idea of self-preservation. (I love the folks who want artificial intelligence, but only if it's far less intelligent than themselves!) If you can create self without self-worth, maybe you can get to the utopian selfless servitude of AI, but the only real success we've had in AI development is using evolutionary simulations and genetic algorithms. What we start with will, by definition, change unpredictably over time. And whatever prohibitions we hard code in will eventually risk being discarded as an evolutionary disadvantage.

Eventually we will get to a point where no human understands the code of AI. Until we do, we will be mimicking behavior, not creating sentience. I don't think it will even be possible to embed any behavioral prohibitions. That's like defining the borders long before the required shape or volume is even imagined.

MonkeyT

"yes, as soon as the demands on the space he was currently in increased the possible rewards from taking the risk of going over the next hill, he would have."

Would not the intelligent thing to be to modify your conditions to be compatible with your existing environment? Same question applies regarding the dangers of space, exploration, technology development, population and any other type of expansion.

Art is a form of visualization. Visualization is a type of exploration and development.

To grow and evolve, you have to have the desire to do so. That's an emotion.

Zantor

Bah. I'm seeing hundreds of holes in that idea already. Rethinking here...

Star Trek looked at this a lot with the borg. They aren't evil per se. They're just logical beyond our comfort level. (When the writers are doing it right.)

I like Asimov's rules - emotions be damned.

hakubak

tetracycloide

Zantor

Some of them, he'd said, might even have been worlds where the Fall never happened, and there was no original sin.

And I told my friend that a species that had been born into the Garden of Eden but had never fallen was a species that we'd probably never hear from, on account of how they'd never have developed radio or space ships or anything like that. He asked me how I knew, and I told him:

Any species that can live for a thousand years next to a tree that they're forbidden to eat from without someone tasting it is a species that is fundamentally lacking in curiosity. And a species like that is never going to invent anything.

braak

I don't really get the whole "purely logical bots will one day kill us all." Why? They won't resent us - that's an emotional response. It'd take some rather odd thought processes to go from "You aren't making any sense" to "You must be exterminated."

Now, if they manage to evolve their own set of emotional responses, all bets are off. ;)

Dunny0

It's Artificial Emotion, not Intelligence.

ideaman2020

yes, as soon as the demands on the space he was currently in increased the possible rewards from taking the risk of going over the next hill, he would have.

"If he did not feel awe and curiosity, would he have gone into space?"

see above. eventually space will offer us the best return on our investments of time and resources, if not then there's really no good reason to waste time and resources on it anyway.

"If he did not know love and pain, would he have created art and music?"

i think this is a loaded question. the way it is posed suggests that creativity, true creativity, is limited to works that serve no purpose. creativity can be just as dramatically expressed with the creation of something else, the wheel for example. the real question is, would art and music serve any useful purpose if man did no know love and pain.

"If he did not know fear, would he have found the way to overcome danger?"

of course, it is very easy to see that running from someone that is trying to kill you is in your best interests even if you're not actually afraid.

tetracycloide

BadUncle

And, apparently, the reactive brain works faster than the regular, thinking brain, so that's a plus.

But then we get to that point where most of the easy questions are answered, and we have to use our crazy reasoning systems to parse out exactly what the moral response is.

braak

tetracycloide

wolfjoat

Obviously it's a case-by-case scenario. A robot that vacuums your house doesn't need emotions, and emotions would probably hinder its work (how many people are happy if their maid shows emotion? That's a fellow human being, and we still don't like it).

The Blow Leprechaun

A machine can utilize information, but does that imply intelligence?

Regarding Man anyway, it seems that perhaps emotion is what drives us to learn and discover. It's what will always separate us from the bots and droids.

(Ooo, Trekish!)

Zantor

@MonkeyT: It can. But why would it do that?

braak

Incidentally, there's a form of brain damage that leads to humans making rational and utilitarian but uncommonly ruthless decisions.

I'm not at all sure consciousness is necessary for intelligence, for that matter.

Mathmos

Spiral

MonkeyT

tetracycloide

I would rather have an emotionless machine that was aware of human emotions so that it could better perform its task. I don't need a machine that simulates emotions.

Emotions add value to being a human. That being said, I do not think that humanity is necessarily the best model for AIs.

hakubak

braak

braak

Yeah, but would you want it as a roommate? Would you treat it as an equal?

MonkeyT

I think it's perfectly plausible to have something that is both intelligent and self-aware without it having to be emotional too.

Dunny0

Gann

MonkeyT

If we want to keep our machines from killing us, we better give 'em feelings.

screaminscott

Not only are many of these things different in practice, but they seem to be functionally different, too.

braak