Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[HauntingPoem]

So I thought (correctly or incorrectly) that we needed a place to discuss/hypothesize/collect/examine everything and anything we know/think/hope about the friendly (mostly) A.I. that live alongside humanity in the QC universe. I for one am fascinated by the way Jeph integrates and writes the different A.I. characters. I also love the recent discussion on A.I. rights and discrimination they enjoy and are subject to respectively.

Feel free to chime in.

[T]

I think autonomous AIs like AnthroPCs are a small fraction of the singularity AI. That's why in the singularity comic it was mentioned like it was an individual. The collectivity of the AIs form the singularity.

This also explain why technology isn't advancing absurdly fast. The singularity learns about the world by having many smaller portions of itself to observe, interact and relate with society and the world. They are all independent individuals and at the same time they are one.

[Is it cold in here?]

A few of us tried to put all the scraps of knowledge together in one place: http://questionablecontent.wikia.com/wiki/AnthroPC

[pwhodges]

I for one am fascinated by the way Jeff integrates and writes the different A.I. characters.

Please remember to spell Jeph's name correctly; it's only polite.

[Is it cold in here?]

There don't seem to be any super-high-IQ AIs. Once they have enough human-equivalent processing power to interact with humans, any further improvements go into processing trillions of inputs and maybe other things for which there is no human equivalent.

Does that mean there's a fundamental dead end to human-style intelligence, or just that there's no reason to create genius AIs?

[HauntingPoem]

I for one am fascinated by the way Jeff integrates and writes the different A.I. characters.

Please remember to spell Jeph's name correctly; it's only polite.

Ahh! You are completely right. Can't believe I wrote that, been reading QC for years!

[themacnut]

There don't seem to be any super-high-IQ AIs. Once they have enough human-equivalent processing power to interact with humans, any further improvements go into processing trillions of inputs and maybe other things for which there is no human equivalent.

Does that mean there's a fundamental dead end to human-style intelligence, or just that there's no reason to create genius AIs?

Could just be that no super-smart AI has made itself known yet; possibly because either such an intellect hasn't arisen, or it may be hiding it's true capabilities to keep humans from freaking out and acting against it (or trying to anyway...).

[Carl-E]

*sigh*

Didn't Station deal with Hanner's psychological issues successfully while running an entire space lab?  I think we're at a level of human meta analysis that no other human could achieve for the first, and the natural speed and multitasking of a huge computer for the second. 

I think they may well dumb it down a bit for their human companions - the way we interact with a dog or small child. 

Of course, my dog's pretty damn smart.     

[FunkyTuba]

There don't seem to be any super-high-IQ AIs.

[citation needed]

What have we seen that indicates IQ of any sort in AIs?
 
All AIs would (presumably) have access to any conceivable online source publicly available (or more) so I doubt access to information (that is, Crystallized Intelligence) is a particular differentiating factor in AI IQs.

That leaves us with (if we're going for simplicity here... just a starting place) Fluid Intelligence or the ability to use logic and reasoning to solve problems in novel situations. As has been pointed out, Station seems to be able to keep a lot of plates spinning. It takes a certain kind of intelligence to (almost) pull off the kind heist May did. Pintsize seems to have maneuvered himself into a pretty comfy gig and finds ways to get access to as much information as he wants to pursue his, other, interests.

Just kinda spitballing here... if it were my universe I'd expect that comparing AI intelligence would need new parameters to do the measuring with.

I also think that AIs in general might hide or not use some significant part of their capabilities in order to pass as more like the people (read humans) they want to interact with. We've seen (or I might be projecting that) people get taken aback by Momo's rapid information retrieval and whatnot... it might be more comfortable for AIs to "dumb themselves down a bit" just to stay in sync with humans, much like Station when he "got drunk". Edit: just noticed that Carl-E made this point already

[Is it cold in here?]

Didn't Station deal with Hanner's psychological issues successfully while running an entire space lab?

Come to think of it, he succeeded where all the best human therapists her parents could afford had failed.

Running the station is more one of the "trillions of inputs" things, qualitatively different from human sentience.

[FunkyTuba]

It seemed to me as though the major component of that was the ability to deliver empathy and care without getting in the way of her phobia of hugs and germs and other people. It was his lack of a body that was primarily important.

That said, I think he might also have an advantage over a human therapist in that his thought processes or approach to sentience was more like hers? So he could also sympathize with her in ways the normal therapists couldn't?

[FunkyTuba]

Just noticed something else...
Station: big computer to operate in, lots of duties, lots of capability
Momo: Seemed (to me) to take on more when she moved to the new bigger more capable chassis
Pintsize/Winslow: still companions but we kind of haven't seen the same depth of character from them as the others.
Toaster: "I make bread FUN!"

Questions: Does processing power have an effect on that Fluid Intelligence I was describing earlier? If you put the toaster AI in station, would it be able to do the same things as Station?

[T]

Another thing to point out is that just because someone is intelligent doesn't means that this person is productive. One could be a genius and spend al his time playing games.

Also Pintsize might be way smarter than he sounds most of the time.
http://questionablecontent.net/view.php?comic=2332

[Is it cold in here?]

It seemed to me as though the major component of that was the ability to deliver empathy and care without getting in the way of her phobia of hugs and germs and other people. It was his lack of a body that was primarily important.

That said, I think he might also have an advantage over a human therapist in that his thought processes or approach to sentience was more like hers? So he could also sympathize with her in ways the normal therapists couldn't?

Wow, hadn't thought of that.

Simply being available 24/7 must have been a huge help. Not even a dog can be that steady a presence.

This strip was my starting point for thinking about why AIs spend processing power on non-human activities rather than upgrading their human-interface subroutines: http://www.questionablecontent.net/view.php?comic=2285

[Sidhekin]

Also relevant: How Pintsize and Winslow react to memory upgrades.  (Too lazy to hunt down the specific comic, but you'll remember when Hanners had some spare MB  for Pintsize ...)

[jwhouk]

Yeah, 512 MB makes Winslow too chipper.

Hey, the old APC models may have intentionally been made that way to reduce the need for excessive computing power.

[Carl-E]

Reminds me of "Your Plastic Pal that's fun to be with" ((c) Sirius Cybernetics Corp.) more than a true companion. 

I don't think it was Momo's personality or mental capabilities that changed with the new chassis, but rather her physical ability to (finally) do useful things. 

Which makes me wonder why they didn't put May into a smaller, more chibi (and so potentially less damage producing) chassis.  I'm guessing it's part of her rehabilitation program, so she can "become a useful, productive member of society". 

It seems that AI fashions and styles are developing at a rather rapid pace in QCverse.  The old ones (like Pintsize, winslow and  and PT410x) are more and more rare.  Momo showed up soon after, but she's a specialized model (from Japan!).  After the practice boyfriend - clearly a prototype that hannerdad got his hands on, or even designed - you get the shopgirls and the larger Momo. 

It's sorta like the jump from the flipphones, Blackberrys and sliding keypad bricks that everyone was using ten years ago to iPhones and Galaxys.  A not-quite-quantum change in the tech that leaves us with a weird mix of the two levels in use.  Pintsize's LAN party was all smaller APC's like him. 

I wonder how much of the change is tied to the AI rights laws.  Which one propagated the other?  Did getting rights make the larger more modern chassis more popular among AI's and their owners, or did the larger chassis necessitate the rights laws?  People being people, I'd bet the second. 

[jwhouk]

I'd suspect the latter, too. It may have been that the older, smaller APC chassis was intentional to allay the fears of humans over a "robot uprising." As time went on, the larger chassis may have become more acceptable as tastes changed and as it was proven robots weren't "out to destroy the world".

At least, not that we know about...

[HauntingPoem]

Well, aside from very vengeful vespas that is...

[jwhouk]

Do you get the feeling that there's some anime character in the QC-verse - a "rock-em, sock-em" robot - that looks amazingly like Pintsize?

[Is it cold in here?]

Deathbot 9000 as well. and laser-equipped small ones.

[T]

Do you get the feeling that there's some anime character in the QC-verse - a "rock-em, sock-em" robot - that looks amazingly like Pintsize?

we already had some anime character cameos. We saw a girl that looked like Professor Shinonome at the space station and Jeph is a fan of Nichijou. http://www.questionablecontent.net/view.php?comic=2143

I have the impression that Nano also made a cameo but I can't really remember when.

[Carl-E]

Well, aside from very vengeful vespas that is...

The vespa didn't care - it was only following orders! 

[HauntingPoem]

Well as far as I know no A.I. companion has to do anything their human counterpart tells them at all. Or am I wrong?

[hedgie]

Even before the AI civil rights thing, and treated more like unusually intelligent pets, they were pretty much like cats.  The human may think they're in charge, but no.  The cat does what it wants to anyhow.

[Is it cold in here?]

Jeph has said it's complete free will.

[Zwammy]

Factors that I wonder about is just how each individual Anthro PC (or forklift, submarine, toaster, etc.) becomes a self-aware AI. The singularity seems based on a particular computer becoming intelligent enough to cross an event horizon into self-awareness, but can you really mass produce AI's like that?

Does an AI compiler spend its time converting a template program into basic AI's, which are inserted into a crčche server until they learn enough to decide how they want to spend their time?

I realize this is a comic, and some artistic license will be taken - but realistically, it doesn't seem like a Pintsize-type robot can house enough hardware to function as he seems capable, and that includes an AI-class intelligence. Especially if 512mb causes problems, since the idea of Momo using most of her computing power to maintain her self-awareness would seem to indicate more ram can only help. The idea that a bigger chassis (with accompanying hardware upgrades) helps increase actual AI capability is a good hypothesis, but how small of a program can function as a self-aware entity? Circuit miniaturization has come a long way, and nano-tech capability would be a factor, but wouldn't things be a lot more futuristic? Station (both the AI and the physical environment) seem to be an exception in modern life, but in the QC world should be more routine in peoples' day to day lives.

Also, Pintsize getting brutalized repeatedly (admittedly deserving of it as he is), rewiring his Sentience matrix, and other unforeseen issues might cause real problems in maintaining enough unbroken logic circuits to maintain AI capability. Overheating? Baaaaad.

Just stuff that runs through my head sometimes.

[T]

My hypothesis is that the singularity isn't based on a single computer but one the collectivity of the AIs.

So a lot of the processing power from companion AIs actually is dedicated to the singularity collective while the remain power is used by the individual chasis.

Each individual AI would be a fragment of the singularity. The singularity as an entity have access to the individual AIs but the individual AIs don't have a direct access to the singularity.

[Mr_Rose]

Momo explained that the first AI was an emergent phenomenon from a certain configuration of hardware and software and that no-one expected it. The next ones were evidently experiments with the same sorts of gear in order to determine repeatability, and eventually someone figured out combos to make it happen on lots of different substrates, but up until the singularity I bet it was pretty old-school trial and error stuff without a fundamentally accurate understanding of what was going on with any precision.

[Carl-E]

Each individual AI would be a fragment of the singularity. The singularity as an entity have access to the individual AIs but the individual AIs don't have a direct access to the singularity.

This bothers me.  As a scientist, accurate usage of correct terminology is ridiculously important. 

The singularity is an event, not an entity/item/intelligence.  It is the event of an artificial "intelligence" becoming a true, sentient intelligence. 

Whether the mechanism is "cloud based" or not, it is apparently independent of computing  power.  But the AI's aren't tapping into a singularity - their existence is the singularity! 

[T]

Yes, it is an event, but it isn't when AI becomes true sentient.
"The technological singularity, or simply the singularity, is a hypothetical moment in time when artificial intelligence, human biological enhancement, or brain-computer interfaces will have progressed to the point of a greater-than-human intelligence, radically changing civilization, and perhaps human nature."

You could have a true sentient AI and still don't reach Singularity.

[jwhouk]

Of course, it may be that once AI's reached "the singularity", they started acting like humans.

And as a result, they stopped acting like computers or robots.

[T]

It should be possible for AIs to act like humans before the singularity.

And singularity could also happen and AIs never act like humans. We might have accomplished a lot of things but we aren't necessarily the most efficient thing.

[Carl-E]

OK, forgive me for going mathematical on the thread, but I think it's a little necessary for what I want to get across...

So in math, a singular point (usually on a function or other mathematical object) is a point at which the item in question is not well-behaved.  It can be a point where a differentiable function isn't, a value where an algebraic function is undefined, a solution to a differential equation that isn't unique at a point, a matrix that has no inverse, ... there are lots of versions, but they all orbit around the same idea, "We have no way of dealing with this because there's no way to know exactly what's going on". 

Now, if you think of a function of time, the one-way variable, and there seems to be a singular point ahead - change (slope) is speeding up remarkably, with no signs of moderation and slowing - you may think you're approaching a singular point on the graph (a vertical asymptote of some sort). 



The thing is, when you're trapped on the function's graph, there's no way to see ahead to find out what's on the other side of the singularity.  Or even how you're going to get past it...

Now, here's the thing - it's pretty well accepted that most change happens exponentially.  Computer processing speed, population changes, human behavioral change, adoption of new technologies, learning curves - there are limits, things like saturation points, but most changes have an exponential base. 

Exponential change does happen faster and faster, that's the whole idea behind it - doubling times, half lives and all that.  But it's not singular. 



Is it sustainable?  Maybe.  And as you look at it from one side, it sure seems that it should have a singularity.  A point at which it heads to infinity. 

But it doesn't. 

Now, things will merge - they already are, look at the progress with mental control of artificial limbs - and again, that change after a merge point will happen exponentially to the point where human augmentation may become quite common.  And we've seen incredible growth in the simulation of intelligence by artificial means.  The two of those may even merge in some way.  it's the stuff of science fiction, sure, but there's a lot of science fiction that becomes true, often before predictions, because predictions tend to be linear (we're linear thinkers), and the real change tends towards the exponential. 

Doesn't mean there's going to be a singularity, though. 

But there's another type of singularity that I think may apply, the non-differentiable sort.  That is, a point where there's a sudden change in direction.  The change that we see happening more and more quickly can all of a sudden do something else - level off, get suddenly steeper (or less steep), or even take a sudden jump.  And, when two paths merge, there's often a sudden change - when technology meets art, when a medical advance changes lifespan or quality, and when a sudden flash of insight goes beyond current knowledge.  Again, the results are (at best) unpredictable.  But it's not the "infinite point" that most people think of or expect when they talk about a singularity. 

Well, sorry for the wall-o-text, but I think it's important to remember where these words we use come from.  I find that it helps make sense of the ideas behind them.   Otherwise, we're just strawmannin'. 

[T]

I like this strawman comic.

[Redball]

The thing is, when you're trapped on the function's graph, there's no way to see ahead to find out what's on the other side of the singularity.  Or even how you're going to get past it...

That made me think of Flatland. I haven't read it since high school. I don't recall that it dealt with being trapped on a graph, but in a college depression a few years later, I wrote some verse that touched on it.

[HauntingPoem]

Thinking about the 1-D place in flatland messes with my head. Also It rains lines in flatland, and the women were lines. Ergo would it not rain women in flatland?!

[Carl-E]

...which immediately made me think of "It's Raining Men"...

Warning: your post has caused this thread to come dangerously close to derailment.  Please review your post and actually think about what you're doing! 

Yeah, yeah. 

[T]

...which immediately made me think of "It's Raining Men"...

Warning: your post has caused this thread to come dangerously close to derailment.  Please review your post and actually think about what you're doing! 

Yeah, yeah.

The last 5 posts already had nothing to do with the topic.

[HauntingPoem]

...which immediately made me think of "It's Raining Men"...

Warning: your post has caused this thread to come dangerously close to derailment.  Please review your post and actually think about what you're doing! 

Yeah, yeah.

The last 5 posts already had nothing to do with the topic.

Alright let us return to the discussion of A.I. What do we know of their emotions. Can A.I. have relationships the same way humans do? If so do they or can they enter emotional relationships or create some kind of inter-personal bond?

[Is it cold in here?]

They certainly form friendships with humans and with each other.

[Zebediah]

Pintsize and Winslow are clearly friends. Also Winslow and Momo. Pintsize and Momo, not so much.

Also, Pintsize apparently had the robotic equivalent of a sexual relationship with another anthroPC - which didn't work out so well.

[HauntingPoem]

I believe I misrepresented my meaning. What I meant to ask is have we ever seen any A.I. become, for the lack of a better term, married/partners/lifelong companions? Or do they not have an interest in forming relationships with each other in the same way?

P.S. I could just be rambling.... 

[FunkyTuba]

Not sure about the lifelong part, but Momo and Pintsize are in some kind of long-term companionate contract with Marigold and Marten. (don't have the relevant comics off the top of my head).

They also chose to do so. Some AIs do not.

I think we haven't seen AI to AI marriages simply because Jeph has chosen not to tell that part of the story yet.

[T]

I believe I misrepresented my meaning. What I meant to ask is have we ever seen any A.I. become, for the lack of a better term, married/partners/lifelong companions? Or do they not have an interest in forming relationships with each other in the same way?

P.S. I could just be rambling.... 

I'm pretty sure somewhere someone already married an AI.
http://questionablecontent.net/view.php?comic=1533
http://www.questionablecontent.net/1658

[Is it cold in here?]

Jeph said in a Q&A once that AnthroPCs were capable of romantic love, potentially with humans, but he's so determined to avoid talking about robot sex that he's not pursued the idea.

[hedgie]

Jeph said in a Q&A once that AnthroPCs were capable of romantic love, potentially with humans, but he's so determined to avoid talking about robot sex that he's not pursued the idea.

I suspect that it'd end up something like this:

[Carl-E]

http://questionablecontent.net/view.php?comic=1533

I misremembered that comic.  Never noticed the second foot on Momo, I always had the mental impression that she was straddling him rather than sitting side-saddle on his lap...

 

And, in the newspost portion of the rickroll comic, we have the following; 

No one is quite sure who decided it would be useful for artificial intelligences to posess libidos, but it is generally agreed that it would be more trouble than it is worth to remove it. Besides, the horny little buggers would revolt.

Now, libidos are not a requirement of romantic love, but they can certainly help.  And what would an AI find attractive?  They can swap chassis easily enough - so it's probably more personality based.  For example, the shop girl (Charlotte?) who had been a nuclear sub - did what she find appealing change with the chassis?  Probably not.  Does she ever visit the crew members she housed in her previous job to catch up on old times?  Was she attached to her captain? 

I think I'm verging on fanfic with some of this, but you get the idea. 

[T]

And what would an AI find attractive?

In Pintsize case I would say ANYTHING!!!

[Storel]

And what would an AI find attractive?

In Pintsize case I would say ANYTHING!!!

Well, anything with B00bZ...