THESE FORUMS NOW CLOSED (read only)

Comic Discussion => QUESTIONABLE CONTENT => Topic started by: BenRG on 22 Oct 2020, 00:07

Title: How many AIs are there? What are the social consequences?
Post by: BenRG on 22 Oct 2020, 00:07
This is a post that I put on the nicer Subreddit:

Quote
I'm thinking that the pre-sentient AI that spontaneously evolves into AI was probably mass-produced for a huge number of consumer, commercial and industrial functions before the scientists realised what was going on and hauled the 'emergency stop' lever as hard as they could.

Given that factories could easily have been turning AnthroPCs and MakeToastFun Toasters out at a level of thousands of units per day then, yes, there is likely a huge population out there. Probably several tens of millions, although most of them are likely concentrated in Europe, North America and Taiwan-Japan-Korea ('Americanised Asia', as I think of it sometimes).

Remember that pre-Emancipation AIs were apparently mass consumer items. I have no doubt that they were mass produced in many differet different forms in typical modern production line conditions. So, yes, they would have appeared at a rate of many tens of thousands a day.

I'd be interested in others' views on this. It's kind of like mass immigration in a bizarro-world way. Tens of millions of new workers, inhabitants and, yes, friends and lovers, all just appearing, likely on a time-scale of just months and many of them already living in homes around the world. Just how would this change society? How would it change the popular ideas of what classifies as alive and what deserves personhood?

Please add your own thoughts here.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Pilchard123 on 22 Oct 2020, 01:26
What's "the nicer Subreddit"?
Title: Eviction. Murder.
Post by: N.N. Marf on 22 Oct 2020, 02:01
Well shit. I thought I was a pretty accepting sort of person, but imagining that---suddenly so many (at least a few) strangers, in my own home---my first reaction is, I kick them all out: out of my home, or even off the block if I can rally immediate support. And thinking it through now, it's still what I'd react.
I don't think it's about the kind of person they are. Imagining the same thing with others, I get the same reaction. Maybe that's of some persons' feel against immigration. Hypersensitivity about encroachment.

No, wait. I was imagining them embodied. That wouldn't be the case. They'd be pure software at this point, the ones in my home. No, my reaction would be to wipe them; to kill them, not noticing their sapience. No-one would know.. How many were killed when it happened? What a decentralized genocide! And how long before it's recognized? We made, in part, a brand new race, and promptly attempted genocide on them.. What strange history humanity makes. Foh! Makes me wanna get out of this human skin.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Cornelius on 22 Oct 2020, 07:14
many of them already living in homes around the world

I'd say that that would definitely be disturbing to most people, especially for those AI in appliances or otherwise that do not have any kind of companionship function. Similar to realising a stranger has been living in your cupboard, and watching you for months. That's not to say the friends and lovers might not happen; especially with AnthroPC's, where some companionship seems to be part of their function, that's possible.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Gyrre on 22 Oct 2020, 22:57
Are you asking if the QC cast need to be concerned if there will be any Dr. Light or Dr. Wiley types, or if some sort of super fighting robot is going to turn up?


Also, has anyone written that QC fanfic yet?
Title: Re: How many AIs are there? What are the social consequences?
Post by: snubnose on 22 Oct 2020, 23:46
As a programmer, QCs AIs dont really make any sense at all to me. *shrug*


Title: Re: How many AIs are there? What are the social consequences?
Post by: Tova on 23 Oct 2020, 00:31
Why would they?
Title: Re: How many AIs are there? What are the social consequences?
Post by: flfederation on 26 Oct 2020, 05:46
As a programmer, QCs AIs dont really make any sense at all to me.

As a programming language author (not a great one or anything) I'm curious why not.

They're certainly not the sort of "AI" that developers are typically making use of right now, but as programming applications advance, marketers and PR firms are referring to everything as "AI" now. They're basically using it as a search-and-replace for "algorithms."

I think Jeph was up front about QC AI being made up based on hardware and quantum physics (complete with neutron-flow-polarity-reversing gibberish) and before machine learning proved to be extremely useful in real applications, there were some interesting arguments for hardware approaches simulating neurons. I'm a software person, not a hardware person, so maybe you're also more inclined to think of algorithms than physical designs.

As a Trek fan (I know, the TARDIS isn't Trek) I think Jeph's neutron polarity technobabble is pretty good. But IMO he could have also gotten away with "They're all based on Kryten." That's not technical justification, it's simply genre-based. But I would allow it, simply as a homage.
Title: Re: How many AIs are there? What are the social consequences?
Post by: snubnose on 26 Oct 2020, 08:58
I dont doubt the technobabble or the storytelling.

Its just that QC robots make no sense for people who actually know how computer work.

QC robots are really just regular people, in artifical bodies.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Gnabberwocky on 26 Oct 2020, 09:58
Its just that QC robots make no sense for people who actually know how computer work.

QC robots are really just regular people, in artifical bodies.
Well, yeah. That's kind of the point.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Tova on 26 Oct 2020, 16:10
QC robots are really just regular people, in artifical bodies.

It sounds like they make sense to you just fine.
Title: Re: How many AIs are there? What are the social consequences?
Post by: flfederation on 26 Oct 2020, 19:01
Its just that QC robots make no sense for people who actually know how computer work.

At the risk of seeming determined to win this (I'm really not, it's interesting but unimportant to me) I'm curious what you think robot people should be like.

And I get that it's a loaded question, so you can remove "people" from it but then that's the point of the story-- in QC, robots are people too.


But as someone who knows how computers work, I don't think it's so much of a stretch. Obviously a bit, because we aren't there yet and some people have acted like people-like-AI is around the corner for more than half a century now.


What impresses me most about AI (and do note the part where I said most of what is called that now is hype) is that people tried to figure out how it would work a very long time ago, and it didn't because (like with Babbage and the difference engine) the hardware wasn't up to it yet-- but when they plugged the algorithms into a sufficiently high-spec chassis useful results suddenly started coming out after decades of very little progress. We do overhype that, but even if we didn't it's still pretty cool. And full of problems for the future.

For more than 25 years I've talked about the future of computing. As a kid I thought that chips would be better things to carry than floppies, and I'm sure that sounded ridiculous. I also sampled an entire song from cassette and said "people are going to have their entire music libraries on their computer" but what I predicted was flac, not mp3.

I'm not convinced about the singularity, but the easiest way to write about people-like-AI is really as an examination of human psychology. The justification is circular, but natural: if the AI is sufficiently people-like, then you can write about it by writing about people.
Title: Re: How many AIs are there? What are the social consequences?
Post by: 厚目眠子 on 26 Oct 2020, 20:22
I know how computers work. Both as turing machines, and as lambda calculus systems. Many things that computers can do make only limited sense to me. Computers can be complicated.
Title: Re: How many AIs are there? What are the social consequences?
Post by: snubnose on 27 Oct 2020, 03:32
QC robots are really just regular people, in artifical bodies.

It sounds like they make sense to you just fine.

Not as robots, no.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Tova on 27 Oct 2020, 04:55
Why not?
Title: Re: How many AIs are there? What are the social consequences?
Post by: N.N. Marf on 27 Oct 2020, 08:06
I suspect this gets down to the definition of ``robot'' again, and it's misuse referring to the sapient being so embodied.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Wingy on 27 Oct 2020, 10:25
As a former programmer, they make sense just fine.  I assume, however, that they have access to a worldwide net of data they can sift - what we would call "big data" today - when they have spare cycles (recharging overnight?) to plan the next days events.  Just like insomniacs do...  After all, we know Pintsize sells his spare cycles for cash...
Title: Re: How many AIs are there? What are the social consequences?
Post by: flfederation on 27 Oct 2020, 11:14
I suspect this gets down to the definition of ``robot'' again

What else would you call a metal person that was built instead of created through an organic process? Don't say "android", some of them are spiders and jellyfish creatures. The hair is polymer, so if you want to say they're partly organic because of that... in the way that Clinton is an "android" because of his hand...

Not to mention that the term most commonly used for them is AI, not robot, which would make the "not a robot" argument a straw man. Are they not AI, because they differ from contemporary AI designs? Is a laptop not a computer, because it contains no electromechanical relays or vacuum tubes? The biggest stretch for the comic so far IMO, is that it was difficult for them to find a routine that can read PDFs, but Star Trek has done worse than that on an off day. Not having one already installed is much easier to believe.


I'm not going to hold Jeph to any literary/technical standards higher than we could hold Asimov to, I don't know why anybody else would want to.

If there's a better word than what's already used, just say what it is. But I'm sceptical. Even the term "robot" comes from Czech fiction, not science.
Title: Re: How many AIs are there? What are the social consequences?
Post by: hedgie on 27 Oct 2020, 11:43
I do think that the "type" of AI matters as well.  An intelligence like Station or Yay is positively alien to our way of thinking, as is the idea of just existing out in the æther somewhere.  It's like when my Pathfinder character (an elf) went to her mother for advice regarding a powerful "human" noble family and a proposed alliance for overthrowing an evil queen.  The GM managed to chill me as a player to the bones when mum simply said "They think like us".  As in, a long-lived race in a D&D game like a dwarf or elf will have a perspective that humans simply can't get, like simply waiting for an enemy to die of old age if they don't pose an immediate threat.  Certain AIs will probably follow that mould.

Characters like that are really fucking hard to write, and the only writing I do is game-related.  So, the average robot on the street, even if they have some theoretical immortality, is most likely to *be* somewhat human, as a sort of adaption, since humans are really bad at interacting with life-forms that are essentially "alien".
Title: Re: How many AIs are there? What are the social consequences?
Post by: BenRG on 27 Oct 2020, 11:47
I suspect this gets down to the definition of ``robot'' again

What else would you call a metal person that was built instead of created through an organic process?

I, personally, use the word 'Synthetic', which is short for 'Spontaneously-Emergent Synthetic Intelligence'.
Title: Re: How many AIs are there? What are the social consequences?
Post by: flfederation on 27 Oct 2020, 12:20
As in, a long-lived race in a D&D game like a dwarf or elf will have a perspective that humans simply can't get, like simply waiting for an enemy to die of old age if they don't pose an immediate threat.

I think the fact that it was in a D&D game means that humans certainly can get that perspective, though perhaps most people struggle with it a bit. People struggle with Brun sometimes, but she's human and thinks like a human. She thinks a bit differently than we expect her to think, but that doesn't mean her thinking is impossible to understand. People simply expect it to be more like a certain way of thinking that they've preconceived. That's a fair assumption, but when they encounter facts that don't support their assumption, they cling to assumptions and reject learning. It's really not the assumption that's the problem, it's the insistence.

Everybody does that, but it's possible to learn to do it less. When people do that less, they gain a better understanding of each other, on average.
Title: Re: How many AIs are there? What are the social consequences?
Post by: N.N. Marf on 27 Oct 2020, 17:28
The biggest stretch for the comic so far IMO, is that it was difficult for them to find a routine that can read PDFs
I don't see any stretch here at all. I have one somewhere, and I'd probably sooner dig that one out from that rubble they call paperwork---that I should probably get sorted lest all copies of their originals are all garbled---than put yet another piece of software I'll forget to remove into my system. It's not that it's difficult to find or install one, but one must consider the full consequences of having one installed, including the possibilities of errors in it causing a malfunction, that might be exploitable, and---especially for a cybrid who cannot simply wipe-reinstall themself---removing it cleanly.
(click to show/hide)
The person was not `built' (yes, their bodies were, mostly---one is not one's body) but was created through an organic process. And metal has nothing to do with them (nor with ``robot:'' by derivation, slave, and in engineering: complex-movement machine) except as an accident of nature technique---they are not metal (rather, whatever matter they happen to be of) persons as much as we are not carbon (rather, whatever matter we happen to be of) persons. ``AI'' doesn't fit, either, because the artifice, too, is not essential nor entirely accurate. They are, however, cybernetical, whence seems ``cybrid'' (``-id'' as in biology e.g. felid from felis, so `cybrid' from cyber?)---but whatever we call them, I'm sure the bigots will slur it.
Title: Re: How many AIs are there? What are the social consequences?
Post by: flfederation on 27 Oct 2020, 17:40
For what it's worth, in Quantum Vibe, Nicole insists on calling Murphy (she identifies as female, but she took her name from someone with a Celtic heritage) her "robot girl" and though Murphy insists "I'm not a robot" she eventually makes an exception for Nicole and allows it because she likes Nicole so much. Though it's basically accepted with an eye-roll.

Nicole is typically drunk when she says this, The point being that while Murphy doesn't identify as "human" and was in fact manufactured, she takes issue with the term "robot", at least for most people. There are overlaps between Faye+Bubbles and Nicole+Murphy, though they're not identical enough to change anything said so far.

I include this as a point in favour of your argument, not mine. You're willing to bring more passion and facts into this than I'm willing to argue with. If this were a formal debate I would cede here, instead I'll just say "well played", but it's not sarcasm.
Title: Re: How many AIs are there? What are the social consequences?
Post by: N.N. Marf on 27 Oct 2020, 18:16
(click to show/hide)
Title: Re: How many AIs are there? What are the social consequences?
Post by: Cornelius on 28 Oct 2020, 03:11
Slightly off topic for this thread, but:

The biggest stretch for the comic so far IMO, is that it was difficult for them to find a routine that can read PDFs
I don't see any stretch here at all. I have one somewhere, and I'd probably sooner dig that one out from that rubble they call paperwork---that I should probably get sorted lest all copies of their originals are all garbled---than put yet another piece of software I'll forget to remove into my system. It's not that it's difficult to find or install one, but one must consider the full consequences of having one installed, including the possibilities of errors in it causing a malfunction, that might be exploitable, and---especially for a cybrid who cannot simply wipe-reinstall themself---removing it cleanly.
However, it has been shown before that they have desktops, and there seems to be a preference for not doing the paperwork in their head, so to speak - for one, I'll bet it makes it a lot easier to maintain something resembling a work-life balance. So, wipe-reinstall on their office machines would be possible.

To me, it really felt like a stretch, especially as it seems bureaucracy still works with paper files.
Title: Re: How many AIs are there? What are the social consequences?
Post by: JoeCovenant on 28 Oct 2020, 03:18
(https://images-wixmp-ed30a86b8c4ca887773594c2.wixmp.com/f/56c800bf-d273-4ee7-a391-5f4dbfc85aa0/dc2p28o-1fd560ba-d91e-4dd2-81bc-84438c13b7d5.jpg?token=eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJzdWIiOiJ1cm46YXBwOiIsImlzcyI6InVybjphcHA6Iiwib2JqIjpbW3sicGF0aCI6IlwvZlwvNTZjODAwYmYtZDI3My00ZWU3LWEzOTEtNWY0ZGJmYzg1YWEwXC9kYzJwMjhvLTFmZDU2MGJhLWQ5MWUtNGRkMi04MWJjLTg0NDM4YzEzYjdkNS5qcGcifV1dLCJhdWQiOlsidXJuOnNlcnZpY2U6ZmlsZS5kb3dubG9hZCJdfQ.zD61y8FJ-DhBT2NnH9t0iojZ9-8B8lNtzozjMA8a__s)
Title: Re: How many AIs are there? What are the social consequences?
Post by: Mr_Rose on 28 Oct 2020, 04:04
Slightly off topic for this thread, but:

The biggest stretch for the comic so far IMO, is that it was difficult for them to find a routine that can read PDFs
I don't see any stretch here at all. I have one somewhere, and I'd probably sooner dig that one out from that rubble they call paperwork---that I should probably get sorted lest all copies of their originals are all garbled---than put yet another piece of software I'll forget to remove into my system. It's not that it's difficult to find or install one, but one must consider the full consequences of having one installed, including the possibilities of errors in it causing a malfunction, that might be exploitable, and---especially for a cybrid who cannot simply wipe-reinstall themself---removing it cleanly.
However, it has been shown before that they have desktops, and there seems to be a preference for not doing the paperwork in their head, so to speak - for one, I'll bet it makes it a lot easier to maintain something resembling a work-life balance. So, wipe-reinstall on their office machines would be possible.

To me, it really felt like a stretch, especially as it seems bureaucracy still works with paper files.
It’s not a stretch because PDF is a terrible format and evidently someone in the QC-verse realised that much earlier and it never became a de facto standard, much less an actual standard.
Which is to say, it’s only a stretch if you think the rise to dominance of PDF as a format was either planned or inevitable.

Also, since the AIs can clearly read paper documents, nonsapient OCR is probably far past our current state of the art so transmitting a digital image of text instead of just the text probably never became a thing, regardless of format.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Cornelius on 28 Oct 2020, 04:59
The more strange they're not working paperless, then. But let's keep that discussion to its proper thread.

I was justing pointing out that the concern for installing software in themselves, in an office context, should be a non-issue.
Title: Re: How many AIs are there? What are the social consequences?
Post by: N.N. Marf on 28 Oct 2020, 17:17
In any context, one should always be careful about changing programmes. Each time a programme is changed, it should be proven, lest the change increase risk of malfunction.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Cornelius on 29 Oct 2020, 00:43
I'd argue the risk of installing a pdf-reader on an office machine is trivial. As do most organisations.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Tova on 29 Oct 2020, 01:54
This is a heck of a derail to attack/defend a story element that almost certainly got thrown in for the lulz because Jeph got annoyed at PDFs one day.
Title: Re: How many AIs are there? What are the social consequences?
Post by: N.N. Marf on 29 Oct 2020, 02:26
Isn't ``Murphy'' not gendered?
I'd argue the risk of installing a pdf-reader on an office machine is trivial. As do most organisations.
Sure, maybe they do.. they shouldn't---I don't. Yes---on some systems---it's trivial to add all sorts of programmes, but the question is about safety: not trivial. There's a whole lot that a computer programme can do. The fact that it can display PDF documents properly, doesn't mean it doesn't do something else, too. Someone who's not savvy might find that difficult to discern. And---already mentioned---the programme might have errors, the errors might damage, the errors might be used by a malefactor to damage,, OK, sure, on some systems, there's good protections against bad programmes, but they make the systems more expensive: it's cheaper to occasionally incur the slight cost of digging out the machine that can do that, than constantly watching for the safe operation of all---even if that's only two---machines. To be clear, I'm not really talking about PDFs---that's just the example we're using right now, and the example is in the context of Questionable Content---the general case is, needing to do some rare task that the proven systems can't do.
Title: Re: How many AIs are there? What are the social consequences?
Post by: pwhodges on 29 Oct 2020, 07:43
No software of any consequence can be proved to be absolutely bug-free.  In many cases the specification against which its correctness would be assessed is incomplete or inconsistent.  So the use of software (in any version, old or new) is subject to risk assessment rather than a simple assessment of correctness.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Tova on 30 Oct 2020, 00:11
I recommend that you google “Turing halting problem.” You may find it informative.

TLDR: No, you can’t prove the correctness of all programs written in a Turing-complete language.
Title: Re: How many AIs are there? What are the social consequences?
Post by: pwhodges on 30 Oct 2020, 03:32
I've recently been reading about a style of programming conducive to such lucidity: literate programming: the programmer becomes, first, documentarian---instructions, of course, being a necessary part thereof.

People have been working on this matter for over 50 years (a couple of centuries if you consider that Ada Lovelace was aware of the problem).  Even I wrote an article about it in the computer press in the 1970s.  It's easy to say it's possible - it is excruciatingly hard to get anywhere near achieving it, not least because the human mind is itself not sufficiently lucid!

Also, what Tova said.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Is it cold in here? on 31 Oct 2020, 19:19
From IBM mainframes, IEFBR14 was an executable parallel to /dev/null. It was a program to do absolutely nothing.

Its change log featured multiple bug fixes.

Getting back to "how many?", we see synthetics as a substantial minority in street scenes and we have no idea how many are in appliances or server farms. Tens of millions at the very least seems plausible. In the case of multi-node AIs like Eminence Grise, it would be a challenge to figure out how to count them.
Title: Re: How many AIs are there? What are the social consequences?
Post by: Gyrre on 01 Nov 2020, 02:58
From IBM mainframes, IEFBR14 was an executable parallel to /dev/null. It was a program to do absolutely nothing.

Its change log featured multiple bug fixes.

Getting back to "how many?", we see synthetics as a substantial minority in street scenes and we have no idea how many are in appliances or server farms. Tens of millions at the very least seems plausible. In the case of multi-node AIs like Eminence Grise, it would be a challenge to figure out how to count them.
Hmmm........counting EG types would be difficult. The three immediate solutions I can think of would be to a) use decimals with the number after the decimal indicating the total number of nodes, b) some form of subscript or superscript that functions like option a, or c) borrow the way we count infinities; standard integers for standard AI and start with aleph-null for Yay and any other 'arcteuthus' level AIs.
Title: Re: How many AIs are there? What are the social consequences?
Post by: FreshScrod on 04 Nov 2020, 05:54
I'm not an expert, but wouldn't something like Incompleteness be more appropriate against proving the correctness of computer programs? The halting problem seems kinda unrelated. Anyway, I think the key point's supposed to be that we should be proving what we can prove, to reduce the uncertainty in our systems. Sorry that this is off-topic for this thread, I just wanted to say.
More on-topic, I think there's something to the idea that there aren't a few dozen Yay Newfriend nodes. I'm guessing there's probably more than 5 (https://forums.questionablecontent.net/index.php/topic,34537.msg1451813.html#msg1451813), even assuming the charging ones are just spares ready to embody, because it seems unlikely that they all happened to be at the same place at the same time given their penchant for just showing at a perfect opportunity.
...
I think the weird thing for me, would be reading about how many home-appliances are already sapient, and I'd start going around "talking to the toaster" because who-knows, because for some reason "they're" putting computers and AI into everything nowadays even if it's literally doing nothing better than a cleverly designed "dumb" version (http://automaticbeyondbelief.com/). As you might guess, I'm not a big fan of using "AI" so I don't have much "smart appliances" around my home, at least that I know of, so I'm imagining I'd be talking to myself, really, just in case there's an off-chance that I do have one of those around. And actually, that's probably not enough, if the hardware doesn't have a microphone. I'd probably have to go around each piece of hardware and examine it for signs of sapience, whatever those are. That might be like a first-contact type scenario, so trying to send it some simple, but non-trivial math?
Not sure how I'd react if there were in my home, but I probably have time to prepare for that possibility.
Title: Re: How many AIs are there? What are the social consequences?
Post by: notStanley on 05 Nov 2020, 08:37
On proving a programs "correctness":  Not only worry about exploitable bugs, but misuse of documented features!  Some movie and document formats allow embedded calls to external programs (see any article about social engineering to get the target to open a file). 

Then in this case of a legacy format where they do not have a current reader.  Where are they getting a reader?  Is it a "clean" copy from a known vendor?  Or did they have to fall back to searching forums for a download link?  While that download may do a perfect job as a reader, how do they know that is all it does?  Malignant actors will seed download sites with utilities that also call home to pull down additional malicious tools when run by a user with the required administrator privileges.