Questionable Content 14: "I Deserve A Fancy Butt Emblem."

[eee]

Eliot's very good at this.

[georgie_leech]

I'll third that. The times where I've been having an episode... well, the times when I've had someone like Eliot around have gone significantly better than when they haven't been around.

[mythmonster2]

That's a neat touch, splitting the speechbubble like that. And yeah, Elliot's a good guy.

[Felius]

For now, while I don't really trust QC to tackle the topic of dysphoria well, it has at least started reasonably well. Only real complain is Melon. Her brand of taking over the scene and creating comic relief in a way that makes everything about her feels ridiculously inappropriate for a serious topic like this. Hell, this very comic, it ends not with Roko's issue but Roko telling Elliot to let Melon have this...

It's seems a very common issue with Melon. She's "quirky" and "random" in those ways that even people who like her like in a rather patronizing manner that don't really treat her as an equal. Which isn't too bad normally, but it does mean that, without any intention of such, everything ends being about her, not about the rest of he world.

[Shadow of the Sun]

Melon said something sensible.

Guys I'm afraid.

[Vinyadan]

I had a deja vu. http://egscomics.com/comic/qa8-08

[eee]

This is an interesting insight into Roko's mind. Things need to go into their place.

[Wraith]

This is an interesting insight into Roko's mind. Things need to go into their place.

Interesting that you saw it that way. I read it and immediately thought that it was another indicator of her highly-integrated mind/body depicting her as being very "human" compared to other AIs, who tend to have very minimalistic requirements.

Momo would sleep in a drawer, before she got a new body. Pintsize apparently has a box of dildos and other sex-stuff stashed somewhere, but no other possessions. Melon has "Art" written on the wall in place of an actual picture. Even Spooky-Bot, as rich and powerful as it is, keeps its sleeping bodies in upright glass "clone tanks" rather than in a bed, or under a blanket or something.
Meanwhile, Roko knits for fun and feels an urge to have redundant but otherwise very human trappings all around her. I'm actually starting to wonder if Jeph isn't going for a transgender-equivalent story, but actually looking to delve more into a transhuman plot as Roko (and other AIs?) start being shown as having more and more human-like quirks.

[halfeye]

I'm actually starting to wonder if Jeph isn't going for a transgender-equivalent story, but actually looking to delve more into a transhuman plot as Roko (and other AIs?) start being shown as having more and more human-like quirks.

Since human quirks include such things as schizophrenia and down's syndrome, I really, really hope not.

[Lethologica]

That seems uncharitable to human quirks. But since Jeph's post-Singularity AIs haven't seemed particularly inhuman (with a couple exceptions), this idea doesn't seem like a rich source of plot for QC.

[Kato]

Jeph seems to think and all-knowing big brother AI is funny. I wonder if he felt the same way about one in real life.

Also, Melon can be cute.

[Max_Killjoy]

Jeph seems to think and all-knowing big brother AI is funny. I wonder if he felt the same way about one in real life.

Yeah, the full confirmation that it's always listening in, that AIs have absolutely no privacy at all... makes me want Station to find that apartment and every other place that AI has hardware, and hit them all with the orbital railguns simultaneously.

[Ibrinar]

They really need better encryption (or security fixes) unless she has physical surveillance everywhere.

[Traab]

They really need better encryption (or security fixes) unless she has physical surveillance everywhere.

And how do you develop better security when spookybot is already everywhere and aware of what you are doing? As soon as you develop new security, spookybot has a counter in place. Its like putting a padlock on the gate after the horse has escaped. Its too late to do any good.

[Ibrinar]

That is not really how security works, sure if she has inserted actual backdoors to guarantee future access then it will be hard to get rid of her but just knowing security measures does not negate them. As in your horse example if the horse is still inside and someone just goes in since it in unsecured them seeing you install a padlock doesn't circumvent the padlock.

[georgie_leech]

That is not really how security works, sure if she has inserted actual backdoors to guarantee future access then it will be hard to get rid of her but just knowing security measures does not negate them. As in your horse example if the horse is still inside and someone just goes in since it in unsecured them seeing you install a padlock doesn't circumvent the padlock.

Unless the security involved is "this doesn't actually exist in anyway shape or form," it's possible to gain access to it. To paraphrase the local Exargh, the point of a door instead of a wall is that it's supposed to open to let people in. Security is designed to only let certain people in. Thus, pretending to be someone allowed in remains a classic way to circumvent security, and programs aren't magic. Watching someone program the way to get past their security is a fantastic way to know how to bypass it.

To further stretch the metaphor, the horse is so far out of the gate that it has ascended to a higher plane of existence and can be wherever it wants at any time.

[Vinyadan]

The problem with putting limits on Squidbot is that she seems to be impervious to them, at least with the current technology. The stuff in SoldierBot's mind was supposed to be uncrackable without spending impossible (not "realistic", as station put it) amounts of time, and she did it in the shortest time. And maybe she simply bypassed the encryption, and, if that's the case, it's hard to imagine how to stop her.

[Ibrinar]

Security through obscurity is an often criticized concept for a reason. Would you say linux can never be as secure as windows because linux is open source?

But lets be concrete and ignore this specific case. Say an attacker cracked encryption and now gets to listen to everything they do to make the encryption better. How does that help the attacker? The biggest encryption concept are publicly available, a specific implementation might have made an error and seeing that could help finding that error. But not every implementation will introduce extra weaknesses.

Now I suppose the ai in QC have no reason to broadcast everything they say so encryption is probably not the culprit. But ais are not remote controlled so there is no reason why the outside has to have access to everything they do so this probably isn't a case where someones credentials got stolen and you can't make new ones since she is watching. So it is likely an unintended way to get info. (Well likely the author has no concrete idea how it gets in but in story I mean.) That can be fixed, if you are a super AI and have the code I guess as long as a vulnerability exists you will find it but it isn't actually impossible for there not to be one it is just hard. (Not counting the vulnerability of login systems of people with access giving you access when they shouldn't, but that is probably not the case here unless AI are obligated to have a backdoor or something.)

If you think her watching makes better security impossible please state a concrete attack scenario? (Personally I believe it is a case of "super AI= can break any system at will" which often occurs in ficton, but that isn't really an in story reason.)

[georgie_leech]

Given that we've seen SpookyBot influence other robots without appearing to do any physical tinkering or direct programming, they're currently at hamdwavium levels of capability. With that in mind, adding "unless you are SpookyBot" remotely to any given kind of security program hardly seems infeasible, or crashing any computer program that attempts to interfere, or doing complex quantum computations to instantly know how to break any form of encryption.

Remember how this forum has discussed how the AI Singularity hasn't seemed to have had much societal impact? SpookyBot is decent evidence that it's because they don't want to, as oppose to any inherent limitation.

Gonna go look up some in comic citations...

On what SpookyBot is

On SpookyBot's programming abilities

On SpookyBot's ability to apply said programs without apparent need conventional access to code

[Max_Killjoy]

Part of what we don't know is how Spookybot is observing.

Physical surveillance devices can be removed from an area if they're found.

If it's direct access, then Roko et al should just be able to turn off any built-in wireless networking they have, and confirm it's off via external independent checks.

If it's something else, then we can't really propose how to block it because we don't know.


I have to wonder if Jeph has thought this out, or is just making the jokes and then backfilling when he has to think about it.

[Traab]

My main point was that its like putting on a padlock when spookybot is there to make his own copy of the key. He sees you putting in security, he sees you developing it, therefore he can see how to bypass/ignore/crack it. Considering this is a being who is likely absurdly faster at thinking/computing/whatever than anyone else, as soon as you start to develop something to block them, spooky is working on a way around it even faster. For all we know there is something in the very base code of AI that allows spooky access that cant be changed without basically starting over from scratch. And yes, we are basically guessing as spookybot works by *mumble mumble* and not by any real rules.

[Mechalich]

Part of what we don't know is how Spookybot is observing.

Physical surveillance devices can be removed from an area if they're found.

If it's direct access, then Roko et al should just be able to turn off any built-in wireless networking they have, and confirm it's off via external independent checks.

If it's something else, then we can't really propose how to block it because we don't know.

Indeed, and we do know there are at least some limits on Spookybot's surveillance capabilities simply because they were not aware of the Bubbles/Corpse Witch situation until Hannelore asked Station about it and Station asked a bunch of other high-powered AIs.

Spookybot apparently developed an interest in Roko following the conclusion of that particular case, for their own inscrutable reasons, and chose to expand resources keeping tabs on her ever since. Nothing in that implies Big Brother-level surveillance, it just means that they are stalking her. That's not surprising actually, as they are on record as saying that they don't consider really any ethical principle other than the sanctity of the mind.

[Kish]

SoldierBot's mind

This is a kind of consistency I hadn't noticed in you before.

[Beelzebub1111]

Is the raccoon dismemberment supposed to be charming or funny? Because that's straight up sociopathic. I'm not endeared, i'm terrified

[georgie_leech]

Because if there's one thing the vague implication that SpookyBot has one of the Great Old Ones as a pet, it's that they're charming and funny.

Have you never read humour into what Pintsize does? Because this is intended to be the same sort of crosses-the-line-twice kind of humour.

[Willie the Duck]

Okay, last panel seems to be shifting art style again, but otherwise a perfectly reasonable Friday cliffhanger.

[Wraith]

If it's something else, then we can't really propose how to block it because we don't know.

I think this is part of the big problem with SpookyBot, because "something else" could very well be "direct infiltration of another AI's brain".

Jeph makes it look funny by, when talking indirectly to SpookyBot, an AI looks up into the air as if addressing a secret hidden camera in the light socket. We also know, however, that all AIs are inherently connected to the wifi and that SpookyBot is a super-Super Computer who can control multiple bodies at the same time.

The "hidden camera" is a joke. SpookyBot certainly doesn't need it, and I think a strong implication is that It is already inside other peoples' heads, looking through their eyes and with only Its own vague "moral code" against compelling other sentient beings that prevents her from taking more active participation. And we know for a fact that this is a conscious choice that it can override whenever it wants to, just as it did to Corpse Witch.

That's horrific, even if only partially true and not even including the part where it openly admits to struggling with the concepts of right and wrong.

[Vinyadan]

Today's comic could also have ended with Squidbot saying "I'm watching". (Assuming she can watch humans through, I don't know, cellphones or (smart) lightbulbs).

[Mith]

Is the raccoon dismemberment supposed to be charming or funny? Because that's straight up sociopathic. I'm not endeared, i'm terrified

Are you talking about the special trough joke, or the fact that presumably one of their dogs brought in a raccoon they hunted down and killed? I mean, it can still be horrific, but I have a similar reaction to my cat sneaking any of his kills into the house.

[Max_Killjoy]

I think this is part of the big problem with SpookyBot, because "something else" could very well be "direct infiltration of another AI's brain".

Jeph makes it look funny by, when talking indirectly to SpookyBot, an AI looks up into the air as if addressing a secret hidden camera in the light socket. We also know, however, that all AIs are inherently connected to the wifi and that SpookyBot is a super-Super Computer who can control multiple bodies at the same time.

The "hidden camera" is a joke. SpookyBot certainly doesn't need it, and I think a strong implication is that It is already inside other peoples' heads, looking through their eyes and with only Its own vague "moral code" against compelling other sentient beings that prevents her from taking more active participation. And we know for a fact that this is a conscious choice that it can override whenever it wants to, just as it did to Corpse Witch.

That's horrific, even if only partially true and not even including the part where it openly admits to struggling with the concepts of right and wrong.

If I were an AI in that setting, the first thing I'd do is get disconnected from all wireless... as in have all the hardware for it physically removed.