11-03-2020, 05:07 PM | #11 | |
Join Date: Nov 2004
Location: Niagara, Canada
|
Re: Building a "Telepathic" AI?
Quote:
Fluff-wise, almost every AI has a completely unique architecture, meaning that each one is going to count as a different 'species' in terms of what affects them mentally. I anticipate all sorts of AI-vs-AI conflict in the background, only some of which spills over enough to be noticeable by near-baseline humanity. I'm probably not going to try to model their Red Queen races (working as hard as possible just to maintain parity) with GURPS; it mostly feels more like plot-device parts of the narrative than anything I need to worry about humans interacting with. ... Though maybe I should try digging up that old GURPS Cthulhu book, just to see if there's anything relevant about eldritch inhuman entities with incomprehensible powers that treat humans as near-forgotten disposable pawns.
__________________
Thank you for your time, -- DataPacRat "Then again, maybe I'm wrong." |
|
11-04-2020, 11:17 AM | #12 |
Join Date: Nov 2004
Location: Niagara, Canada
|
Re: Building a "Telepathic" AI?
I'm considering bringing in computer/program Complexity rules here, by letting these AIs shove these advantages into "skill set" computer programs, ala Transhuman Space; in the 4e adaptation, the base rate is C3 for up to 2 CP, C4 for 3-4, C5 for 5-8, and so on to C9 for 65-128. Of course, the AIs involved would need a better Modular Ability than comes standard with THS AIs (30 CP for 3 slots of 2 CP each, only for skills and languages, and with -3 to the slot's rolls in sudden emergencies or combat). So a 42-CP version of Intuition would be a C8 program, while a much less-effective 27-CP version would only be C7.
This gives me an easy handle on letting the AIs bargain with each other, exchanging particularly useful sub-programs in exchange for resources, or the like. (Other than one AI simply eating another AI whole, absorbing all their knowledge, and possible reprogramming them and letting them loose; which still remains an option.) And also gives me an excuse for why an AI might not have the best-possible version, if whoever they traded for it only released a crippleware version. (And gives me another reason to come up with the most Limitation-laden versions of the Advantages that I can think of, to lower the Complexity needed when running lots of them at once, or for AIs with relatively small Modular slots.) And gives me a way to guesstimate how long any given program takes to write, with the New Invention rules on pB473-4. Anyone see a problem with the basic mechanics of this? Are there any supplements with additions or alternatives to GURPS' basic software-writing rules?
__________________
Thank you for your time, -- DataPacRat "Then again, maybe I'm wrong." |
11-04-2020, 06:58 PM | #13 | |
Join Date: Nov 2004
Location: Niagara, Canada
|
Re: Building a "Telepathic" AI?
Quote:
And does anyone have any thoughts on how a critical failure during the Prototype stage would lead to 2d damage to the programmer, and having to pay the Facilities cost again? (The best I've got so far is spending excessive amounts of cash on the local TL's version of Amazon Web Services for neural-net training; and/or a death-march overtime crash-project leading to excessive fatigue, which either leads to medical issues, or tired individuals physically tripping over themselves, bumping into walls, accidentally triggering halon systems, failing to cook the chicken in their lunch, or the like. ... I'm having a bit more trouble figuring out the equivalent when the programmers are AIs, though.)
__________________
Thank you for your time, -- DataPacRat "Then again, maybe I'm wrong." |
|
11-04-2020, 07:26 PM | #14 | |
Join Date: Feb 2005
Location: Berkeley, CA
|
Re: Building a "Telepathic" AI?
Quote:
Unless the software is controlling a robot or something similar, ignore it. |
|
11-05-2020, 09:22 AM | #15 |
Join Date: Jan 2014
|
Re: Building a "Telepathic" AI?
|
Thread Tools | |
Display Modes | |
|
|