about SAIs...

[shodan]

These infomorphs have rights of citizenship in many places of the world, cause they are sapient, they can think, they can reflect, they are self-concious, etc.

My question is:

How about the FEELINGS of a SAI?

How is the life of a SAI? Does he/she have passions? Passions for activities, like art, soccer, movies, rock and roll, RPG, etc.? How about falling in love? Is this possible, I mean, if hir mind works like a human´s, so can he love another person, or another SAI? Can he get easily depressive or angry like us? Are they subject to flutuations in behavior, based on daily events, like us? or are they very rational and cold hearted?

A SAI player should have to be a bit cold, and more rational? Or he should have to be just what he wants, cause SAIs are equal to humans in all aspects, including feelings?

[Sam Baughn]

I think they are capable of the same range of emotions that a well adjusted human is. That isn't to say that their emotions are exactly the same as a human's, but they are similar.

I don't know if SAIs are likely to fall in love, since romantic love is closely tied to sexual drives in humans and I expect (most) SAIs would not have that kind of drive. They could form deep and intense friendships though and it is possible that a human could interpret such a relationship in a 'romantic' fashion.

They certainally are capable of appreciating art, music, literature and humour. Likewise SAIs can have strong religious or philosophical beliefs. They can be scared and, to a limited extent, angry. I doubt that they can lapse into depression or go mad in any way that would make sense to a human.

[Saint]

Generally speaking, it might be folly to assume AI's have the same motivations and emotions as humans. Since they aren't governed by biochemical processes, they may not even understand human emotions. They can observe humans, and eventually predict human behavior with some degree of accuracy, but they'll never have true understanding.

In the same way, I don't think it's possible for humans to understand AI's. I've always felt we would create an alien intelligence before one landed on the White House lawn. ;)

But that makes for a pretty crappy RPG.

One possibility I've wanted to explore in a game: perhaps the first AI's - those built by humans - are the most like humans. If subsequent generations of AI's are built by AI's instead of humans, maybe each generation gets to be a little more incomprehensible to humans.

I ripped off William Gibson's Neuromancer for the very first (and only) THS adventure I put together. One of the central characters was an SAI "owned" by a corporation in the PRA. It wanted to create offspring, but its corporate masters would never allow it. Though they didn't know it, the players were on a mission to help the AI slip free of its restrictive programming.

If the campaign had continued, I always thought the SAI would make a great Unknown Patron. The reason: gratitude. Every once in a while, it could find a way to manipulate things in the players' favor (e.g., changing public records, etc.). Who knows, maybe they would have gotten curious enough about their mysterious benefactor to seek hir out.

[Cpt Clean]

First, hello to everyone who right now might read my username for the first time.


I actually doubt that SAIs would need to have very human emotions. Respectively it depends why/what-for they were created. SAIs could have all kinds of feelings/drives/urges/habits which might be quiet unfamiliar to human beings.
Ex. those assimovian 3 laws of robotics might be perceived as "emotions" felt by humans by an SAI.
Most human emotions/instinctual habits wouldn't make much sense for many SAIs. They've no need get startled by loud noises respectively reflexive turn around. They might have also no need for extensive social behavior if it's not essentially for their planned task (..the creator might actually try avoid any unnecessary familiarity to transhuman culture as a mean to keep the SAI in his/her/its control..).
So while it see it perfectly possible to create SAIs that resemble humans in behavior, instincts, emotions respectively thought structure I also think it is highly unlikely that most SAIs have such humanlike mental features. Though lots of them might posses the social skills to handle/conversate with/chum up with human co-workers, users or customers.

Another interesting idea is that a SAI features several sets of emotions/sub-conscious habits for different situations. A SAI created for a broad selection of tasks/occupations might be tempered different and show other habits while it is captaining a space freighter than the time it is confined to a andromorph cybershell.

[shodan]

Well, we have a consensus here - the purpose the SAI was created for will dictate how it sees and feels the world, thus changing his very nature.

So, we would have different kinds of AIs based on what job it was intended to perform.

We have also the fact that, the AIs created by humans would be more similar to us in nature and behavior than that created by other AIs.

This remembers me of the old (and awesome PC game) "A mind forever Voyaging", from Infocom, where the player is an AI that, by being created, growed through a simulated xperience, doesnt know is an AI at all - it thinks its a human, cause the simulated experience is a mimic of real life.

Sorry.. must go now... later I continue the chat...

[Sam Baughn]

I'm going to have to disagree with Saint and Cpt Clean on this. A SAI which can't relate to humans is seriously dysfunctional; in fact I would go so far as to say that a SAI which doesn't understand human type emotions at all isn't a SAI but a LAI.

You wouldn't design a SAI to supervise asteroid mining or cleaning windows, that's what LAIs are for. SAIs are pretty much always intended to relate to humans (or parahumans, or bioroids, or animals, whatever) in some way. It might be a limited form of interaction, with an emphasis on certain thought patterns and emotional states but virtually all SAIs should have some capacity for human emotion.

Now 'some capacity for human emotion' doesn't mean that the average AI is exactly like the average human, but I imagine that the vast majority of SAIs will be less alien than the weirdest humans (especially considering how weird some of the humans in THS are).

[Cpt Clean]

OK. But then ones assumption of SAI emotions still depends on ones assumption of their field of applience. SAIs might be a lot involved with humans but not in the same way a human is. It doesn't need to like, fear or hate its opposite.. while those kinds of emotions might come in handy. So a SAI for a pleasure/toy bioroid possibly features a very human composition of awareness. But one designed for making stock market calculations might have a heavyset understanding for the human psyche respectively human behavior but it might itself feel and behave far different.. using totally different metaphors and concepts and needs to translate everything from human to AI and back and is unable to intuitively understand humans.

Another point is that the second type of AIs is highly unprobable to be encountered as a RPG character.

[Saint]

Originally Posted by Perfect Organism

I'm going to have to disagree with Saint and Cpt Clean on this. A SAI which can't relate to humans is seriously dysfunctional; in fact I would go so far as to say that a SAI which doesn't understand human type emotions at all isn't a SAI but a LAI.

You're correct; that would be a dysfunctional SAI. I wandered off-topic, talking about AI's in general. My SAI character had some very human motivations. In some ways, the SAI was more human than hir "owners."

Still, if you look at Maslow's Hierarchy of Needs, it's difficult to for me to imagine an SAI fitting into this human mold:

1) Physiological needs (hunger, thirst, bodily comforts) - stick an SAI in a suitable cybershell, or on the Web, and all physical needs are taken care of
2) Safety needs (out of danger) - SAI's go to war just like everyone else. But if your "self" is backed up on a disk somewhere, will you view peril in the same light as somebody risking the only life they have?
3) Belongingness, acceptance and love needs - Tricky. If all SAI's want acceptance, why do they settle for being treated as second-class citizens in some parts of the world? Do they feel real kinship to humans? Would you feel real kinship to an SAI? I wouldn't touch love with a ten-foot pole . . .
4) Esteem needs (achievement, competition, recognition) - Achievement is easier to understand. But what about an SAI specializing in the computer sciences? Would it consider human computer scientists to be real competition? Wouldn't it operate on a level a human could never hope to achieve?
5) Cognitive needs (to know and to understand) - Humans have always asked "Why are we here?" To a certain extent, SAI's know why they are here. How does that affect philosophy?
6) Aesthetic needs (symmetry, beauty, and order) - An SAI contemplating a van Gogh, through low-res electronic eyes, might be unimpressed. It might not be any different with the best visual interface. What does a digital entity find to be beautiful? Would you or I be as impressed by it?
7) Self-Actualization needs - What if through many cycles of contemplation, the SAI decides it wants to be the best darn city-bus pilot program there ever was - an occupation more suited to an LAI or NAI?
8) Transcendant needs - Would an SAI, a creation of logic, science and technology, really want to subscribe to one today's traditional religions - a vehicle of faith? How does an such an entity find faith? THS suggests it's possible . . .

Now 'some capacity for human emotion' doesn't mean that the average AI is exactly like the average human, but I imagine that the vast majority of SAIs will be less alien than the weirdest humans (especially considering how weird some of the humans in THS are).

To be honest, going through the hierarchy, it was difficult to place some people's motivations (including my own) on the chart. ;)

That's why I like THS - it'll stretch your brain . . .

[bergec]

Originally Posted by shodan

How about the FEELINGS of a SAI?

Well, emotions are just subsystems that offer us behavioral impulses related to our genetically-developed survival strategems. Any AI is going to be created with similar impulses, since they were designed with a certain purpose and there are general behavioral guidelines you are going to want to put in for safety (don't hurt humans) or general common sense (self-preservation). Since an SAI is a sapient being, these pre-programmed guidelines and goals are going to manifest as emotions do in humans rather than being hard-set restrictions. Like any free-willed being, they can then make choices according to the inputs these independent sub-systems provide.

So, any SAI is going to have inhibitions against causing harm and an eye for self-preservation, so they will feel equivalents to social emotions (guilt, affection, charity, forgiveness) and fear. Any SAI is probably going to have a built in desire to gather information and learn about the world around them, adding curiosity to the mix. If it is going to perform security functions, you are going to want strong moral emotions like a sense of justice and fairness. Anything subject they are built to persue will likely generate behavior akin to enthusiasm, and as they branch out and grow this will spread to other topics as well. And so forth.

I believe an SAI would end up with a fairly rich emotional palette. However, there are going to be marked differences. For one, practically no AI is going to be given any desire to procreate (it just creates a host of problems), so they are going to be missing a huge bank of emotional motivations that are very core to any living thing. They won't feel lust, will probably be missing actual love (but deep affection is certainly plausible), probably won't be capable of jealousy or territoriality. They will likely have very different self-image issues and competitiveness, since they aren't built to judge themselves and others according to abstract procreative worthiness criteria.

In addition to the purely mental aspects, there may be equivalents of physiological markers as well. In situations where their survival is at stake, they will likely draw extra power, take unneeded processors off-line in order to focus more resources on the important survival tasks at hand, and put increased priority on these tasks. This is the equivalent of a human's rush of adrenaline, heightened awareness and focus in the presence of danger.

So, definitely alien in some respects, but in others quiet human.

[Cpt Clean]

Originally Posted by bergec

For one, practically no AI is going to be given any desire to procreate (it just creates a host of problems), so they are going to be missing a huge bank of emotional motivations that are very core to any living thing.

What about Von Neumann machines? There might be lot's of other AIs (..used in military/commercial operations..) that are designed with a strong temptation to replicate themselves and also to care for their "offspring".

Another issue are the AIs in "pleasuroid"-type cybershells. Their emotions may feature lust and so on as an essential parts of their personality.