THESE FORUMS NOW CLOSED (read only)

Comic Discussion => QUESTIONABLE CONTENT => Topic started by: Morituri on 26 May 2018, 20:42

Title: Do AI in Questionable Content design and assemble their own offspring?
Post by: Morituri on 26 May 2018, 20:42

Obviously the first few generations of robot chassis were designed and built by humans. 

But, in a post-singularity world, it's generally considered likely that most new AI (software and processing hardware) will be designed by AI.  And it will almost certainly be the case that however the construction of chassis happens it will likely be robots doing most of the assembly work - whether that's at a factory, or elsewhere.

It's already been mentioned in-comic that most AI spend their first few years in virtual reality, and get downloaded into a chassis where they can interact with the physical world only once somebody (who, exactly?) is convinced they're sane - or at least, not dangerous.  Melon may be an example of "neither sane nor dangerous" and she's out in the world.

Anyway...  it seems like all the plausible parts are there for AI parents to design the minds of their offspring and then design and assemble bodies for them as well.

And even if that's not the way most AI do it, Faye and Bubbles both know more about the design and assembly of robot chassis than pretty much everybody by this point; they could.  If they wanted to, they could have (build) children - an equal collaboration by both partners, and no outside help from a donor would be required.   

It would almost certainly be considered weird for a human to be in a 'parent' role to a new AI.  Would it also be considered weird by the AI community for the offspring to get a 'child-size' chassis and not spend childhood in Virtual? 



Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: snufflebottoms on 26 May 2018, 20:45
They might be working on that technology!
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: Undrneath on 26 May 2018, 20:56
I would think that the creation of an AI from two AI parents would be the combination of half of each of the parents code much like the biologic counterpart.
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: awgiedawgie on 26 May 2018, 21:06

Obviously the first few generations of robot chassis were designed and built by humans. 

But, in a post-singularity world, it's generally considered likely that most new AI (software and processing hardware) will be designed by AI.  And it will almost certainly be the case that however the construction of chassis happens it will likely be robots doing most of the assembly work - whether that's at a factory, or elsewhere.

It's already been mentioned in-comic that most AI spend their first few years in virtual reality, and get downloaded into a chassis where they can interact with the physical world only once somebody (who, exactly?) is convinced they're sane - or at least, not dangerous.  Melon may be an example of "neither sane nor dangerous" and she's out in the world.

Anyway...  it seems like all the plausible parts are there for AI parents to design the minds of their offspring and then design and assemble bodies for them as well.

And even if that's not the way most AI do it, Faye and Bubbles both know more about the design and assembly of robot chassis than pretty much everybody by this point; they could.  If they wanted to, they could have (build) children - an equal collaboration by both partners, and no outside help from a donor would be required.   

It would almost certainly be considered weird for a human to be in a 'parent' role to a new AI.  Would it also be considered weird by the AI community for the offspring to get a 'child-size' chassis and not spend childhood in Virtual?
I recall seeing somewhere in the comics that the first AI spent a while in VR, but I don’t remember seeing anything about most AIs doing that. Since AIs are as uncertain as humans about how AIs happen, it seems that they would not be any more or less able to create new ones. The process is repeatable... it’s just that no one knows how it works.
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: Is it cold in here? on 26 May 2018, 21:31
Jeph said they spend their early life in a VR creche. There's probably a quote somewhere in the wiki.

The idea that Faye and Bubbles could have a reasonable number of babies and have them be synthetics had not even occurred to me.  Fascinating!

Yes, AIs could do half of each parent, but they're not limited to that and might prefer something more intentional.
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: BenRG on 27 May 2018, 00:46
Regarding the question in the thread title, I would argue that the fact that the Singularity has happened demands that this be the case. What remains unknown is exactly how this happens.

My personal favourite idea is that it actually is a pseudo-biological process. We know that when two Synthetics have their analogue of sex, they network their systems together. I assume that, normally, firewalls are in place to stop too much data merging between the two processes. However, if both participants bring down their fire-walls, their intelligence algorithms splice together, generating a third algorithm that contains aspects of both donors.

What happens next is the analogue of pregnancy. The 'child' algorithm remains on the 'mother's AI drive in a protected partition but able to access all the sensory input from the chassis. This is the equivalent of being in the VR crèche for factory-compiled AIs and is where the algorithm gets to acquire and process sufficient data to emerge into full consciousness. Once this process is complete (a process of a few months, likely), it is uploaded into a chassis. The result is an AI with a very immature personality (think of May, or early Winslow) that is actually the direct offspring of two other intelligences.

Why this idea interests me is Spookybot's jak headset. It should be at least theoretically possible to use that to merge a human's brainwave patterns with an AI algorithm in the same way as two mating AIs would. The resulting AI child would then have elements of a human's intelligence patterns too, a neurological hybrid of human and machine. Trans-human, if you will. As the modern humanoid chassis tend to have pseudo-biological elements, a true human-machine hybrid is at least theoretical possibility from this process.
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: fayelovesbubbles on 27 May 2018, 08:59
Uh...AIs have offspring?
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: ToodleLew on 27 May 2018, 13:37
<Courtroom_voice mode="Perry Mason">

Your honor, I object. The question assumes facts not in evidence ("AIs have offspring", "AIs mate, or otherwise combine to produce offspring", etc.), and calls for speculation from the witness.

I suggest that, if my learned opponents desire to establish the point, that they call Jeph Jacques, or his proxy, to the stand as an expert witness.
</Courtroom_voice>
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: Neko_Ali on 27 May 2018, 13:43
Uh...AIs have offspring?

There has been no sign of that, at all. There have been no parent/child dynamics between AIs shown, talked about or suggested in the comic. Nor any AIs talking about becoming or desiring to become parents. So no, I don't think that happens. We also know that aside from possibly the first true AI, they aren't created. Humans or other AIs don't program new AIs. They just seem to happen....

My personal headcanon is that new AIs can occur spontaneously. Where a complex system becomes self aware or the mingling of other AIs gives rise to the new one. But nobody thinks of the contributing AI as parents because either it's something that can't be forced to happen, and/or it takes a lot more than just two contributing AI, so as a culture they tend to think of new AIs as part of a culture and are cared for as a group instead of a handful of parents taking responsiblity.

Another option is that new AIs are created by those god-level AI that are sometimes mentioned. The ones who eschew any sort of physical body Momo has talked about how their thought process and computing power is well beyond what an embodied AI is because they don't have to spend processing time using and controlling physical bodies. That they are so beyond what even other AIs can understand and they are the source of new AIs.
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: Mr_Rose on 27 May 2018, 14:53
From what I can tell, Canon is that the first true AI emerged from a specific combination of hardware and software, some of which may have been developmental or even experimental, and may have even been intentional.
The first was, however, an isolated case for some time until the examt circumstances to trigger AI development in that specific setup were determined. From there, equivalent setups with less expensive hardware were discovered and eventually commodified.
Until the Singularity, no-one actually understood why the specific combinations worked to create sapience, instead of simple sentience or even just basic responsiveness, and the new knowledge may now be being applied to human sapience.

As for robotics, I imagine that the whole planet was just a few years ahead of us with companies like Boston Dynamics beginning to commercialise products rather than still developing them as is the case in our reality.
In turn one of the AI research teams probably got ahold of one, rigged an interface for their pet mind, and quickly determined that the rapid learning and intuitive reasoning of the AI combined incredibly well with the dynamic environmental awareness of the robot body.
After that, well, people have wanted robot butlers for centuries, and now here they were; potentially at least. The initial commercial success was probably limited because the AIs couldn’t not have personality and the type of person who wants to be the first to own a mechanical slave for the sake of showing off is seldom the type to tolerate backtalk from their “property”…
Still, by the time the idea of the companion matching service gained traction, the cost of the hardware had come down to mass-market band, and the idea of having an AI as a sort of pseudo-pet, rather than a disappointingly wilful minion, became popular.

Which is how Marten and Pintsize came to be associated, just a little while before the strip starts…
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: awgiedawgie on 27 May 2018, 18:23
Uh...AIs have offspring?

There has been no sign of that, at all. There have been no parent/child dynamics between AIs shown, talked about or suggested in the comic. Nor any AIs talking about becoming or desiring to become parents. So no, I don't think that happens. We also know that aside from possibly the first true AI, they aren't created. Humans or other AIs don't program new AIs. They just seem to happen....

My personal headcanon is that new AIs can occur spontaneously. Where a complex system becomes self aware or the mingling of other AIs gives rise to the new one. But nobody thinks of the contributing AI as parents because either it's something that can't be forced to happen, and/or it takes a lot more than just two contributing AI, so as a culture they tend to think of new AIs as part of a culture and are cared for as a group instead of a handful of parents taking responsiblity.

Another option is that new AIs are created by those god-level AI that are sometimes mentioned. The ones who eschew any sort of physical body Momo has talked about how their thought process and computing power is well beyond what an embodied AI is because they don't have to spend processing time using and controlling physical bodies. That they are so beyond what even other AIs can understand and they are the source of new AIs.
What about Hannelore’s Roomba and (I assume) Mrs. Roomba, and their three little Roombinos (and the guys at the Roomba factory commenting about the one with the jet engine trying to reproduce with another one)? Or when Jeremy and Seven were confessing their mutual attraction, and a comment in the strip (or maybe it was Jeph’s subtitle) said something along the lines of “they’re going to have a million robo-babies”? Sorry, I’d give links, but my laptop just puked, so it’s a little more difficult right now.
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: Gyrre on 30 May 2018, 06:01
"Half of each parent" kind of neglects the nurture side of things, doesn't it?

Also, couldn't doing that with coding lead to some really wonky stuff?

 :-o What if Melon is a 2nd generation AI?!?
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: awgiedawgie on 30 May 2018, 08:54
"Half of each parent" kind of neglects the nurture side of things, doesn't it?

Also, couldn't doing that with coding lead to some really wonky stuff?

 :o What if Melon is a 2nd generation AI?!?
But "half of each parent" is the nature side - the physical creation of a new AI. Nurturing would be whatever involvement the "parents" might have in the development of the "child" after that.

The "really wonky stuff" in humans is far more likely when the parents are closely related, for technical reasons that might not apply at all to AIs.

I like to think that Melon is a savant, and she just hasn't found her area of profound intellect yet.
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: Undrneath on 30 May 2018, 10:06
I like to think that Melon is a savant, and she just hasn't found her area of profound intellect yet.

Perhaps poker (http://questionablecontent.net/view.php?comic=3746)
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: awgiedawgie on 30 May 2018, 10:25
I like to think that Melon is a savant, and she just hasn't found her area of profound intellect yet.

Perhaps poker (http://questionablecontent.net/view.php?comic=3746)
Considering that she was eating the cards... I doubt it.
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: Undrneath on 30 May 2018, 11:04
She still had most of the chips though.
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: OldGoat on 30 May 2018, 11:14
<Courtroom_voice mode="Perry Mason">

Your honor, I object. The question assumes facts not in evidence ("AIs have offspring", "AIs mate, or otherwise combine to produce offspring", etc.), and calls for speculation from the witness.

I suggest that, if my learned opponents desire to establish the point, that they call Jeph Jacques, or his proxy, to the stand as an expert witness.
</Courtroom_voice>
Damn Lew, you're good at that!
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: awgiedawgie on 30 May 2018, 13:56
She still had most of the chips though.
But as someone pointed out, all of her chips are different colours from everyone else’s. She may not even be playing the same game. She may think she is, and they may have just given her a bunch of chips to keep her happy.
Title: Re: Do AI in Questionable Content design and assemble their own offspring?
Post by: ToodleLew on 30 May 2018, 16:25
Damn Lew, you're good at that!

Curses, you've pierced my disguise as ToodleLew, and exposed my secret identity.

Well, I'm retired anyway. Maybe I'll go back to my Perry Mason novels ;-)
Nah - the conversation here is more entertaining.