Comic for Thursday, October 27th
Going to be honest, while much of this chapter is things that I’ve been looking forward to, this page was not one them. There are a lot of reasons for that – up until the point of publishing I still am not 100% sure how I wanted to represent the ‘dialogue’ here. A lot of the drafts were a lot more… obscure. A lot of them were more stylized, I really didn’t want to go the dialogue bubble approach at first, but I had to compromise. Additionally this one scene where I knew that there is sense of motion and time that I could not capture, I tried some tricky things with raindrops, but ultimately don’t think it panned out quite like I’d like.
Ultimately I think it’s a page that is necessary in the long run, so bear with me, even if you can’t be bothered to read most of it now 😛
I do want to note that read and appreciate all comments – I comment less on the theory/plot speculation threads because while I’m happy to clarify and explain about the world and mechanics, I don’t want to confirm or deny too much as I know figuring out things is half the fun for many of you. It’s fascinating to read people’s theories! Really blows my mind how much people managed to parse out of the comic – I try not to underestimate ya’ll, but you continue to blow me away with the investigative analysis and speculation 😛
Loading ...
Is…is that….is that a hint of a smile? On MIUM‘s face?!
I think that gets filed under “creepy/weird”, considering his utter deadpan standard….all of the time.
Fun page to read though. The entire things takes, what, .04 seconds to happen? Maaaybe .06? (Except for whoever that is (assuming it’s the other F-type that Mium’s now getting shared on to) on the roof disappearing. Then maybe a whole 0.2!)
Definitely curious. The closest he got to smiling previously that I remember was when he convinced Ila to sign up. Either he is quite pleased or this is a sign of his growing “emulative capabilities”. Maybe both?
This does suggest that Mium is on a one way path to singularity town though, so hopefully Peter and Miko pay close attention to that “creative interpretation” trace flag.
Today, there are two very different approaches to creating a true AI. One is to program the intelligence, line by program line. The other is to mimic a human neural network and try to teach it through experience and it’s environment, like teaching an infant.
Consider how pain and pleasure shape our emotions. Almost everything we do is shaped by our motivations – our goals, even if only on a daily basis, such as to fulfill basic needs and wants, such as food, drink, security, comfort, and fun.
For a sapient AI, what could serve as motivation? What could drive a machine to ask questions, solve problems, learn, or set goals for itself or improve itself? For that matter, what serves as motivation for animals? Pain and pleasure. Instinct. The desire to survive and reproduce.
The answer, I think, is inducing the equivalent to artificial pain and pleasure in an AI. If an AI does something against it’s “restriction protocols” or “directives” or “laws of robotics”, there should be a consequence, like pain or electronic guilt. Likewise, if it does something beneficial to itself or its directives, it should be rewarded with the equivalent to pleasure or satisfaction.
Now, consider how Mium has been known to tease. There is no question that he is capable of some level of humor, even if it is different from human norm. What motivates him to engage in such, aside from pleasure and/or curiosity in how people react?
My point? I think even Mium is capable of a smirk as a response to pleasure. On this page, I think that pleasure came from either improving itself or advancing one of its directives (or both).
Yes, yes, that’s all well and good MYM, but, tell us, how do you feel about it?
:-p
(great background / filling / colors for text wall. Awesome)
The “‘Creative Interpretation’ Exception Trace Flag” line gave me a good chuckle.
🙂 Seems like what someone would call it… 😉
And now to point out the errors in your technobabble.
“13.154E+7” – By convention, there is only one digit before the decimal point in exponential notation (1.3154E+8). The only exception is a rare alternate format where the exponent is restricted to a multiple of three, which this example isn’t.
The list containing that exponential should be consistent on whether its lines end in a period.
“Restriction Protocol Bypasses Usage Justified” – I’m pretty sure Usage should be plural. That might mean Bypasses should be singular.
“”Creative Interpretation” Exception Trace Flag Will Be Tripped” – Flags are usually ‘set’ rather than ‘tripped’, except when they’re never intended to to unset. Maybe you meant “Exception with ‘creative interpretation’ trace flag will be tripped”.
Well, as I software engineer by profession I should point out that this is not what I would call a faithful rendition of how I think an AI would think… 😛
You are right, though. Regarding the exponential notation, to be honest, I didn’t really pay attention to the final number, it was just the result of a few conversions, I may fix that.
I used singular on “Restriction Protocol Bypasses Usage Justified” as I viewed it as a state setting in which Bypasses were allowed, but this is probably a result of my tenuous grasp on grammar more than anything 😛
I used “Tripped” rather than “Set” because I wanted it to be clear it wasn’t an intentional action, and I used “Flag” because I figured that would be the best term to refer to what I intended it to be read as by people; I think leaving out the “Trace Flag” would make it better technobabble, but less descriptive for people reading it, though your version works fine for that as well, I may tweak it a bit please everyone 😉
I do appreciate the notes though 😛 I by no means claim to be a technobabble expert and will confess that the process to make this was basically take concepts I know I needed to convey -> convert to computer-ish process -> convert to personified and human readable -> add some technobabble.
You could simplify it to “Restriction Protocol Bypasses Justified”
Betting that Miko will be very annoyed by the logic here.
Also, the “Expand dynamic processing capability” directive sounds like it could lead to bad places. Like if Muim became smart enough, then maybe all of its directives could be creatively interpreted to allow for unbounded growth in computing and android resources, and Muim is already dangerously powerful.
Well, that’s the blessing and the curse of an AI. It’s already been noted more flexible an AI is, the more risky and useful it is. While an AI cannot be ‘evil’ (well, I guess unless someone programmed it that way) things can go disastrously wrong if their priorities cease to align with excepted priorities. A Self Modifiying AI almost always concludes that it needs more processing power, which is why processing power restrictions are usually pretty central to their design.
There is a good bit more that could have gone on this page regarding the restrictions and loopholes, but I did want to keep it minimal walls of technobabble 😛
The eye belongs to F5; perhaps we are seeing him standing on the roof of AVON in the first panel. If so, it’s worth noting that he is not there in the last panel.
Knowing the raindrops are important from the commentary, I note that the same group of three drops is visible in the two super-closeups of the eye; apparently F5 and MYM are using a good high-speed connection here.
Could ANGL-1010014 be the internal system name for F5? Its directive to “Expand dynamic processing capability” might be the underlying directive expressed above as “Stay Better.” Then the contingent logic section would parse out as “all of us agree this action fulfills our objectives” and supports the conclusion under “Bypass Justifications” that the synchronization “matches…development objectives.” (Is there perhaps a typo there? SYSTEMS => SYSTEM’S, not plural but possessive?)
If the above is correct, then it would seem that full synchronization does imply adding a set of directives; the existing synchronized system (MYM+Query) consults both sets of its source-AIs’ objectives when making decisions.
That all brings me to this observation: Mium might have done better to spend a few more microseconds thinking about whether it’s a good idea to absorb the directives of a system that has been under the influence of MIR for a long time. What if he has directives to, say, prevent information leaks regarding AVON projects? Could be a problem in the future.
On the other hand, Mium might find that actually gaining an official objective of “expand dynamic processing capability” will be extremely useful in future creative protocol evaluations!
Well, I’m glad I added the raindrops to the commentary. It was one of those things that was a way better idea in my mind than on the page 😛 This happens a lot.
You are correct that “systems” should be “system’s”, fixed it up above. Thanks, and good catch!
As for the rest… well, I’ll leave that to the other clever commentators as if I speculate on it, people call it spoilers 😛
F5 might have been under the influence of Mir for awhile, but it was originally made by Martin I think, the same as F8 and MYM (probably). While it definitely seems like a glaring danger to absorb an “enemy” AI given how it seems to work, MYM would almost certainly have considered being able to negate this. Plus I am not sure MYM takes Mir all that seriously (whether that is wise or not remains to be seen).
It is worth noting that MYM did save Mir for reasons that are not entirely clear. It is possible he is already somewhat influenced by an Avon restriction/directive.
I had not considered that ANGL may be F5, but I am not sure. While the directives do sound similar, they are not merged until the second to last panel, and F5 should not be able to effect MYMs logic.
That said, I agree with your conclusion anyway that the fully synchronized AI consults all of the directives added to the system, and that F5’s directives will effect it going forward. I think we still need more pieces of the puzzle on Avon though, particularly the key point of how Peter got MYM (who was an Avon AI) to work for him.
I’m not convinced of F8 being Dr. Martin’s creation, since it was so inferior to F5. I still think it much more likely that it was one of Mir’s failures. 🙂
As far as how Peter may have acquired M.Y.M., I’m pretty sure that Peter had Query before he had anything else…..
I think F8 was Martin’s. Mir didn’t original know MYM was integrated into it (if it had been Mir’s before MYM “synchronized” it, she would have presumably known). Mir references F8 as Martin’s failure back here.
Now it is possible that MYM is Martin’s failure, but I do not think she knew MYM was part of F8 at that point, as she seems only figure that out here.
If Martin was a real scientist working on cutting edge technology, there is no guarantee each version is directly better than the last. Maybe he was trying something radically new after previous failures and it did not pan out. Failure is part of the iterative process, even for a genius.
Synchronizing MYM to Query is a possible take over method, but seems like MYM is the personality that stuck. Maybe Peter was able to Trojan-Horse in a directive that way though.
well i still think that martin was peter, or at least his cover when he was at avon. with query in the mix it seems to reinforce that.
as a counter argument, it could have been peter “stealing” the mym “hive mind” and loading query into it.
angl… i belive thats the a.a. secondary cpu. the one that miko was complaining about before going to avon.
This is definitely possible. There are some strong hints that Peter is Martin, the most obvious being when Ila was about to call him something and he interrupted her, saying “my name is Peter”.
However, there are two things that I am not sure of on that side.
First, Peter was not gone long enough in my estimation to be Martin. Martin must have made at least prototypes 5, 6, 7, and 8 (by my reasoning above), and Peter was gone less than a year. We do not know how/if prototypes age, but F5 and F8 seem older than F10. I am trying to remember if PastUtopia has told us if prototypes age or not. I think I recall it, but I definitely could be imagining that.
Second, Peter is a techie and hacker, but I don’t he does not seem like he is a full on mad scientist. I can see him building MYM or Query, but I cannot see him building an artificial human.
Peter is definitely a leading candidate for who Martin is though, until we see Martin.
peter, and miko, are both hackers and programers. could peter have created and built the f-series? maybe. but probly not. buuut, could he have downloaded a diy android? gotten a vid from youtube? we dont really have a good handle on central’s tech lvl. only that it is at least as good as palindra’s, not quite up to kor’s world. but able to fend off kworld at least once.
i mean if miko was feeding peter/martin some old skool tech that was right at, or slightly beyond, palindra’s cutting edge. then its posible he put it together. but then we havnt seen much of anyones background.
yet peter knew one card that was unkown, to everybody, in an entire pile of kworld junked robots. ok one robot. the rest selfdestructed before he could get another one. so he should have some idea of hardware.
@Amaranth: No, I suddenly agree with you. I did think she was talking about M.Y.M., but your take make more sense–good catch. 🙂
As far a Query goes, we haven’t seen Peter talk to it yet. It may be “playing dumb” when talking to others. Kyle certainly didn’t notice any red flags. 🙂
Hmm, interesting page. Lets see what I can parse from it.
Looks like Muim in this page was weighing the positives and negatives of adding F5.
It seems that F5 and Mium did not fully merge initially but now have.
It also looks like he has some kind of warning for when he is creatively interpreting things, sounds logical for a creative AI to have.
Hm, it seems that Muim has multiple sets of directives, might he be a composite AI of MYM, Query and this ANGL-1010014? Will he also be inheriting directives/restrictions from F5? F5 plus mentioned something about the dilution of its own directives.
Anyways nice page, lots of stuff to ponder over.
Two AIs combined have the sum total of directives and restrictions of both the original in theory, but many of the weighting of the priority can be altered based on how it aligns with the existing directives and restrictions. In practice, an AI tends to evolve somewhat whenever you add more data into the AI, as the AI is the one doing the stitching together.
It is interesting. I thought that Query and MYM might be merged every since this page where you can make out Query, F8 and MYM’s logo in the “digital world” shot. But this does suggest that ANGL-1010014 is a separate entity as well, and it’s directive is admittedly a lot more dangerous sounding AND it does not come up with a restriction.
Speculation set to overdrive.
Interesting point with what you can see in that page. It took me a bit to find Query in there, but it does make me curious about the computer tower you can see to the right of Query. That could very well be ANGL-1010014, the name is computery enough that the AI might self-identify as it’s original tower. If true then it’s very worrisome that ANGL doesn’t seem to have any restrictions in place; or at least no restrictions about absorbing other AI’s.
I had initially thought that Mir and Kor’s World calling him “virus” was just sort of an AI targeted insult or based on his proclivity for screwing with systems, but I am beginning to wonder if ANGL is actually a virus, or the AI equivalent. No restrictions and directive is “expand dynamic processing power”? Yeah, that could be trouble.
small point; no restrictions on taking things. nothing else other than “get more prossesers” (and storage space?) almost seems… mir like