Comic for Tuesday, July 23rd, 2019
…okay, so yes, this is a bit late. In every sense of the word. I should have gone to bed more hours ago than I can currently count.
Anyway, least we have a comic 😀
The digital world is a bit… eh; I had an idea of how I wanted to look, but could not really get that to work, so just fiddled with a few things until I decided good enough. It’s intentionally different then what we have seen before, but with some similarities, and I think that much is still intact in this version even if much of the original idea isn’t really. It’s background that can mostly not be seen, so I probably worried about it too much given that it’s most essentially a… imaginary world? anyway.
When VI Mium notes he was based of off Miko, he isn’t necessarily lying, but that is also not the entire truth – he was adapted from something original made to interface with Miko’s system adapted to interface with Ila’s system. Of course, if that actually has anything to do with him being “a brat”, well, that seems sort of unlikely, given Mium default, though VI Mium does act someone differently.
Anyway, I have to wake up for work in… hmm… about 5 hours so…. going to hit post now and stop rambling.
Loading ...
This comment is a response to comments about self driving cars down thread but it deserves to go to the top.
Past I am sorta shocked that you do not see the issues involved in self driving cars. You have this wonderful comic about General AIs in all their perturbations leading to all the complications they bring about. How do you not see the issues involved in turning control of autos over to an AI.
Who should the AI driver be loyal to and in what priority
the owner of record,
the driver,
the manufacture (the car company),
the passengers,
the person who hooks up the charge cord and changes the tires,
or the government.
keep in mind that these can all be different with different goals.
If the AI witness a crime should it snitch, and at what level of crime, for example speeding, or murder. THINK CAREFULLY ABOUT IT.
Should an automotive AI snitch on an immoral act like adultery for example. THINK CAREFULLY ABOUT IT.
An automotive AI must be able to break some rules or the effect is indistinguishable from malicious compliance.
If you put a high priority on pedestrian safety then pedestrians will bully self driving cars
unless you make jaywalking a serious crime,
or you allow the car to use its superior weight to gently shove pedestrians out of the way in some circumstances.
The above dose not even get into who is liable for the actions of an automotive AI.
Also many people are afraid of self driving cars being made mandatory or effectively so.
The AI in this comic produce all kinds of complications, yes… and yet I’d consider their impact “good” on the whole.
For self-driving cars: are there both ethical and technical complications to work out? Yes, absolutely, but “not figuring them out” shouldn’t be any more acceptable just because a human is in charge of the car instead of an AI. The best solution to “my solution is systemically wrong, beyond random error” should not be “add more random error.” (Of course, computers make different mistakes than humans, but I expect that to be at absolute worst a wash once we get to the point where self-driving cars are road-legal.)
Let’s be honest, though. Most human drivers are going to try to not hit pedestrians too – even when the pedestrian on question is doing something stupid and/or dangerous.
Also, the level of “intelligence” in self driving cars is pretty low.
A Car’s self-driving systems aren’t going to be able to recognize or understand most of what you’ve mentioned. It’d be like trying to bypass a retinal scanner by putting a fingerprint up to it instead of an eyeball.
Current “AI” tech can really only do one thing. They’re more of an “expert system” than “intelligence”. A self-driving car “AI” can only drive itself.
How worried are you about the “AI” running the opposition in single player games? Not very, right?
As for any recordings the systems’ sensors might make … presumably it would be accessible for the owner to review, and accessible to others under certain conditions (ie, if there’s an incident, law enforcement can get warrants if they know which cars they want the information from: likewise, the insurance companies are going to want access to the logs to help determine fault). Probably also some mostly blind access by whomever is responsible for coding the self-driving programming in the event of bugs or other reported errors.
And liability for what the self-driving system does should go to the company responsible for its code.
Apply all of this logic to a smart phone.
So, who owns your smart phone?
The manufacturer
The OS writer
The carrier
The NSA
Law enforcement
Krackers
Porn Sites.
All have more rights and privileges over your phone than you do.
Any one of these groups can turn your phone into a brick at any time.
Any one of these groups can use your phone to “tattle” on you at any time.
So, long before we talk about self-driving cars, we should tackle the legal ramifications of phones.
Of course, all this will be made moot by linux phones.
Giving real power to the people.
Welcome to the MASSIVE pile of issues that actually hampers the development of work toward true AI’s right now. Its too easy to fudge up and fudging up will probably be irreversible. So how do you go about it, when you only get one shot?
Good thing smart cars don’t need true AI’s.
I think that it’s more a shift about how we think about cars. No one really bats an eye at that the majority of the subways in the world are basically fully automated (most of the drivers are ceremonial if they exist, though this is not true everywhere). A car system is a really complicated subway system. It’s more akin to a highly advanced Roomba than Mium. You might think “Well, I don’t want my Roomba’s AI driving a death machine around” until you compare the intelligence in your Roomba to the average driver and realize those are already about the same.
I think questions about who it is loyal to should be treated with the same skepticism as your cellphone. I wouldn’t recommend using your cellphone or selfdriving cars you wouldn’t in theory want large tech companies or the government to know about, but in 99.9% of cases they necessarily don’t care as no one can use that data for much (in an individual sense). It’s like asking if Google should report crimes discussed in gmail – they could, in theory, but they don’t already. Privacy is technology concern, not a self-driving car concern. Questions like reporting in are the same as the question around your cellphone – should spyware be placed on your cellphone? Probably not, but we are going to use them anyway. Spyware wouldn’t be a main functionality of your car, and quite frankly they could already put spyware on a not-self driving modern car as they have computers and lots of cameras, and that’s pretty much all they need to do that much.
In general it will try to avoid crashing as that’s bad for everyone; diverting is almost never the solution to collision – the infamous trolley car problem is mostly a thought experiment. We don’t really want humans swerving or diverting cars either – that causes as many collisions as not (or more). A self driving car would never be programmed to pick which targets are the most valuable, as that doesn’t make sense, it would just try to stop. I mean, I don’t work on self driving cars, but I work on automating large sums of money which is something large corporations are generally less risk tolerant of then human monkey lives, and they are automating all of that because – at the end of the day – automation screws up in a big way less often than human monkeys, and I am fairly confident the programmers working on self-driving cars are at least as good as those working on business software automation.
But the real heart of the matter is that people will absolutely drive cars that might kill them if the chance of it doing that is lower than the chance of the person behind them killing them by not stopping for a stoplight… and right now the chance of the person behind you killing you is fairly high, so the bar for self-driving cars isn’t super high (compared to how good people seem to think it’d have to be).
I am pro-self driving car; but I also use a cellphone and gmail. Various tech companies already know where I am 99% of the time because they know where my cellphone is. They already have a microphone and camera that follow me around constantly. Privacy laws are not unenforcable, they just need to exist first – the solution cannot be “don’t have the means to spy on you” because they already do that; the solution has to be “actually create and enforce laws about it” – while some people are cynical, enforcing that sort of laws is not impossible. Corporations have been regulated in the past, and could be regulated again in the future. Unfortunately (and trying not to get overly political here) the entire collection of my government understands how technology works about as well as a house cat understands laser pointers – they chase the latest and shinest issue, while ignoring everything else about it and have no real understanding of the underlying cause.
Some could argue that we shouldn’t have self-driving cars until we have legislation to regulate more about them, but I don’t think that’s realistic, and I don’t think they move the needle on privacy for most people significantly.
I can absolutely see the merit of requiring self-driving cars in city limits; outside of a city there could be a large parking lot to park driver-not-included cars. The traffic difference between a fully-self driving system and a mixed system would be billions of dollars and millions of hours saved. I certainly can see why some people wouldn’t like it, and I wouldn’t compare it to the people that wanted to keep their horses on the road after cars were popularized as it’s not quite the same thing, but I think the benefits outweigh the cons.
Now I should note that I am not unequivically “pro AI”. While Mium is a very stylized version of AI, there are real concerns with general AI, but that is still a far ways off and will not be realized in the sci-fi sense anytime particularly soon (10+ years); self-driving cars are functionally already here on the other hand, and are not what I would really consider AI anymore than the business software I write is AI. I write software that directly replaces humans, but that’s because a lot of what humans do only looks like it requires decisions because they have imperfect information – a system with more information can solve that problem. Self-driving cars the same way; there aren’t really decisions involved in how to get from A to B in a car if you know all the factors involved.
Of course, talking to someone that’s got rear ended a few days ago about if they think monkeys or computers should drive cars is probably a somewhat biased conclusion… 🙂
On your closing note… it’s probably worth mentioning that I’ve been having bad dreams involving car accidents since 1980, at *least*. It’s hard to say for certain if I’d been having them before that; my early dream journals were very sparse. They no longer exist, so I can’t check them, but I know there was an event in 1980 that caused a dramatic uptick in the frequency of these sorts of dreams.
Hey guys; just want to update – I’m still alive, but we are going to miss… uh… Thursday’s comic… at this point. I will still have a comic for Monday I think. Just behind on stuff. I got rear-ended on Thursday, work has been busy, and just not doing very good keeping productive on things – if it makes you feel better, I’m behind on everything else as much as I’m behind on the comic 🙂
But just wanted to update here so you guys know that I’m still alive (despite someone’s best effort there 😐 #selfdrivingcarswhen; I wasn’t not significantly injured, and even the car is mostly fine, just cosmetic damage I think).
Ultimately all these excuses are just more reasons I need to have a buffer, but buffer trees haven’t been growing recently, so not sure where to find one 😀
I’m glad you’re okay!
Having your car be mostly fine, too, is a bonus, if not an insignificant one.
And… really, if you need time, you can just take it. There is *no* need for you to have a car accident to make your taking the time be okay. Honest!
I just want to keep people in the loop; I know that most people here are very patient, but I also don’t like just vanishing as that’s how a lot of webcomics end, and I don’t want people to think it has ended or is ending.
And it is appreciated.
Doesn’t mean we won’t tease you about having an accident to get some down time!
Hmmm….
void *allocate_comic_buffer(int size) {
void *buffer = malloc(size * sizeof(comic));
return NULL;
}
I think there’s a problem with this code somewhere…
I have, by the way, started to think there’s two kinds of drivers:
1. Drivers who are longing for self-driving cars.
2. Bad drivers.
These two kinds are not mutually exclusive.
There is a special case I’ve thought of that should be pointed out: there are some drivers who basically never encounter other drivers on the road who may not want self-driving cars, but have no reason to think they’re bad at it. That said, trust me, they are. They may be fine at driving where nobody else is around, but their inexperience with dealing with other cars means that they’re fundamentally bad at dealing with that situation. There isn’t room in this comment to express how important experience with multi-car situations is needed to be a good driver in multi-car situations.
I think a lot of people view self-driving cars as taking away their control of their car, as they think they are somehow avoiding accidents. The reality is though that you only need a very small percentage of people to suck at driving to have a lot of accidents where there is nothing much you could do about it (and that percentage may not be very small – people are just not good at reflecting on how good they are at something – most people think most people are bad drivers, but they are a good driver).
Self-driving cars prevent other people from hitting you as much as prevent you from hitting other people, and will get much much better once most cars are self-driving as they can be like “yo, so i was thinking about moving into that spot, you cool?” to other cars on the road instead of communicating like humans do with finger gestures (usually of the middle variety) and the occasional horn blare with a side of cursing (or more often just complete obliviousness to their existence).
TGape, one obvious problem is that your code didn’t start with:
#!/usr/bin/perl
use strict;
use warnings;
Very few C code samples do. That said, I intended my sample to be inserted deep enough in a file that it’s almost a certainty that the file already has #include <stdlib.h>, otherwise the lack of that line is certainly a huge problem. I don’t think that’s the bug I felt the routine had.
I just noticed… Interlude 6 is in the archive before chapter 1. Also, this update doesn’t have a section listing, so it’s in void.
… my filing system can only handle having one update in void. So either this needs to be put in a section, or I need to figure out how to distinguish from .
Should be fixed; the order seems to break everytime I add a new chapter.
Panel 3: why am i even waking on the ground in the digital imaginary worlds now?
I think your comment would be more clear if you pointed out that ‘waking’ should probably be ‘walking’, or something. At least, when I’m looking at somebody’s review of my code, if they just quote a line like that, most of the time I’m just confused.
No, “waking” is correct. She has of late developed a habit of waking on the ground.
willis just added an “i” and a “the”. The former looks like an improvement (modulo case); the latter doesn’t look like an improvement to me. Then again, I’m not a native speaker, so apply appropriate amounts of sodium chloride …
The “the” would change the meaning from “digital imaginary worlds” in general to “the digital imaginary worlds are implicitly a specific place, and I’m waking up on the floor now there in particular” – native speaker for what it’s worth, and I agree that “the” is perhaps not what is intended.
Of course she is waking on the ground. Her useless meatbag body kept throwing itself out of its chair
Would it be too much of a spoiler to get Miko’s swearing at the scientists in panel 6 posted in a comment so we could read the whole thing? It does tend to be amusing.
Just want to make it known that I am loving Miko’s long hair, looks fantastic.
Agreed.
Also looking forward to the strip when new VI Mium is introduced to Peter and things are explained concisely. Hopefully Mium F8, F5, and Ila don’t try to nuke him.
Miko: Dr Mir is crazy, she turned herself into a robot.
(Goes and turns herself into a computer.)
That said I really have to wonder about the definition of “alive” in this comic. Dr Martin has cracks on his skin similar to Atter has at one point, and Atter isn’t alive in the traditional “blood and bones” sense, so does that mean that Dr Martin is also a “demon”? Maybe, if demons are constructs that gained sentience, he is the opposite: a sentient that gained magical construct properties.
If that is true than it begs the question, what is Miko now? AI doesn’t seem to fit, since she isn’t technically “artificial” (ie: manufactured) in origin, but she seems to run off the same type of coding on the same type of systems. Is there even a name for that? The fantasy equivalent would be a “being of pure thought”, but is there even a sci-fi term for it?
I believe cyborg still applies. She still has her meat brain. Its just that she also has a computer brain running most of her thoughts at the moment. Or we COULD call her a whole brain emulation, even though it’s only temporary for her.
There are effectively two Miko’s now. I’m not entirely certain how temporary this is for her. Can she be merged back with her meat-self? Or perhaps there really isn’t a distinction yet, and she’s just still having trouble staying completely synced between what is closer to two hemispheres of her “brain”?
I think “uploaded” is generally used for “… from the brain into a computer”.
Well, infomorph fits.
https://www.orionsarm.com/eg-article/465cd5ad0c747
Upload does as well, as possibly does dividual.