I am looking for stats to use to create an AI. Is (or was) there a book in third edition that gave some stat blocks for one of the AIs that have populated the Matrix at onetime or another?
Or if you have thoughts and theories on what stats they should have that'd be good too.
Thank you for your time.
Full Version: AI stats
It was Matrix maybe?
It wasn't very specific, only saying things like, this AI should be roughly twice as powerful as the PCs, or this one is beyond the pale as far as runners are concerned.
It wasn't very specific, only saying things like, this AI should be roughly twice as powerful as the PCs, or this one is beyond the pale as far as runners are concerned.
Welll..... you take the number 8... and you turn it 90 degrees. And those are your stats for a True AI. All of them.
That's about the size of it.
I'm with Xenith. You give them the same stats they gave Harlequin in his namesake adventures. That was:
Harly is a plot device, he succeeds at whatever he tries.
Harly is a plot device, he succeeds at whatever he tries.
Although that is certainly true in the case of the 'God' (Passion?) AIs like Deus and company, if you are wanting something a little more managable personally I think that I'd treat them as basically self-aware high rated Agents with some of the Technomancer flavor thrown in for added spice... (For example, I'd allow them some form of Threading and what basically ammounts to Submersion...)
well there was the semi-ai programs known as SK (hell if ill attempt at spelling the full name). the interesting thing about those was that they was able to write their own programs depending on what their task was iirc.
in effect, take a agent, give it a point pool (say rating squared or something like that) thats the total rating in programs said agent can deploy at any one time. if it needs a stronger attack program it can maybe lower its browse or analyze (or maybe drop them completly) and increase the attack program based on the points released.
please note that the rules for limiting the effective rating of programs and so on would still apply...
in effect, take a agent, give it a point pool (say rating squared or something like that) thats the total rating in programs said agent can deploy at any one time. if it needs a stronger attack program it can maybe lower its browse or analyze (or maybe drop them completly) and increase the attack program based on the points released.
please note that the rules for limiting the effective rating of programs and so on would still apply...
just as a note on the whole harlequin deus thang. The same was said about the great dragons, but now they're statted out, least ways at a baseline. Sure with a few hundred karma they could raise their stats, but they'd rather spend it initiating which is much cheaper.
SK semi-autonomous knowbot. A nice low level ai which true ai's supposedly evolved from.
SK semi-autonomous knowbot. A nice low level ai which true ai's supposedly evolved from.
Even creatures with God stats can be killed. Just look at Deus and pals the instant before Crash 2.0
And you might not even need another plot device to do so. Trap most of the AI on a node somewhere, destroy the avenues of retreat and then smash the node(s). Instant dead AI.
And you might not even need another plot device to do so. Trap most of the AI on a node somewhere, destroy the avenues of retreat and then smash the node(s). Instant dead AI.
| QUOTE (Xenith) |
| Even creatures with God stats can be killed. Just look at Deus and pals the instant before Crash 2.0 And you might not even need another plot device to do so. Trap most of the AI on a node somewhere, destroy the avenues of retreat and then smash the node(s). Instant dead AI. |
vel there is allways the "death by a 1000 bee stings"...
Here's a simpler or lesser AI. My thoughts go along the lines of using Great Dragon stats as the byline.
Complex Forms
All at Rating 8. (dragons break skill rules so an AI should get to do it to)
Int Passes: 4
Resonance: 12+
Damage Track of 21 or higher.
Firewall of 13
Resonance: 14
System: 13
Biofeedback: 14
Initiative: 27
Can have 26 Sprites bound to him.
Skills:
Cracking Group 8 (once again dragons break skill)
Electronics Group 6 (can use drones
Resonance Group 8
Various Pilot skills for use on drones, If it lacks he right program it can likely steal the proper pilot program
And more, anything it lacks but needs it could likely get a hold of in the form of skill softs.
Now this is how the AI's grew out of control they simply absorbed all the knowledge of the matrix.
Complex Forms
All at Rating 8. (dragons break skill rules so an AI should get to do it to)
Int Passes: 4
Resonance: 12+
Damage Track of 21 or higher.
Firewall of 13
Resonance: 14
System: 13
Biofeedback: 14
Initiative: 27
Can have 26 Sprites bound to him.
Skills:
Cracking Group 8 (once again dragons break skill)
Electronics Group 6 (can use drones
Resonance Group 8
Various Pilot skills for use on drones, If it lacks he right program it can likely steal the proper pilot program
And more, anything it lacks but needs it could likely get a hold of in the form of skill softs.
Now this is how the AI's grew out of control they simply absorbed all the knowledge of the matrix.
Is there evidence of AIs using sprites? If not, I'd be tempted to just use a ridiculously high-rating sprite that was of all types and had all the sprite powers, and just run with that.
| QUOTE (Aaron) |
| Is there evidence of AIs using sprites? If not, I'd be tempted to just use a ridiculously high-rating sprite that was of all types and had all the sprite powers, and just run with that. |
Well when the AI's were around spirtes didn't exist as technomancers didn't exist. Many technomancers are the result of the crash and what the AI's were up to.
| QUOTE (Jack Kain) |
| Well when the AI's were around spirtes didn't exist as technomancers didn't exist. Many technomancers are the result of the crash and what the AI's were up to. |
Weren't there Daemons and Sprites in SR3? From the otaku?
The trouble with AIs is locality. Where exactly are they? In SR3 Deus was centrally located in the Acrology. In Crash 2.0 he was specifically recompiled at a specific time with the stock exchange as the focus.
Now that the Matrix is much more decentralized, the problem becomes how do you pin an AI down? Is the sum of their parts less than the whole? You could probably pose a Schrondiger's Cat dilemma if you tried hard enough.
Now that the Matrix is much more decentralized, the problem becomes how do you pin an AI down? Is the sum of their parts less than the whole? You could probably pose a Schrondiger's Cat dilemma if you tried hard enough.
I would see the AI's moving more like an entity like sprites or agents. They wouldn't attain the god status of before because of the decenteralization but there is nothing to prevent lesser AI's from existing.
Seems to me that AIs need hard coding somehow. Deus and Morgan/ Megaera were always either in a person's head or on a complex host. Mirage's location was more indistinct, but it definitely resided physically on semi-dormant servers. In System Crash it definitely implies that AIs have source hardware, just like metahumans.
| QUOTE (Glayvin34) |
| Seems to me that AIs need hard coding somehow. Deus and Morgan/ Megaera were always either in a person's head or on a complex host. Mirage's location was more indistinct, but it definitely resided physically on semi-dormant servers. In System Crash it definitely implies that AIs have source hardware, just like metahumans. |
There are free willed sprites roaming about. But I'm not talking about Deus or Morgan level AI's. Something smaller, greater then a simply high rating sprite but less then Deus.
SK or daemon by the sound of it. but still, the SK had a home node/host.
but the AIs sounded like they created that from distributed processing.
deus chose the stock exchange because it not only had the capacity to house his home host fully, but because it would allow "him" to upgrade himself. so he put all his eggs in one basket. i guess it shows that even being a AI do not make one a military genius
and where was morgan/meagera housed between the escape from renraku and the capture and dissection? or when "she" again escaped, before returning to battle deus in the arcology and ending up coming along when deus dispersed himself the people?
but in the end, AIs in SR is one of the worlds big unknowns...
but the AIs sounded like they created that from distributed processing.
deus chose the stock exchange because it not only had the capacity to house his home host fully, but because it would allow "him" to upgrade himself. so he put all his eggs in one basket. i guess it shows that even being a AI do not make one a military genius
and where was morgan/meagera housed between the escape from renraku and the capture and dissection? or when "she" again escaped, before returning to battle deus in the arcology and ending up coming along when deus dispersed himself the people?
but in the end, AIs in SR is one of the worlds big unknowns...
Does anyone have Captain Chaos's old stats lying around...?
In order to create an AI, you need three things. First, you need a high-rating autonomous program. SKs work fine but an Agent might do in a pinch. Second, you need hardware capable of running that software, obviously. Third, you need !.
A good ! can really pick up the slack for bad hardware and software.
Compare, for example, Number Five to the T-800. The T-800 is a highly advanced robot from the future, but is ultimately limited by its program despite its superiority. Number Five is a 80s military robot that was built by the lowest bidder (which happened to be Steve Guttenburg). He probably had an Apple II for a brain. Yet, Number Five was hit by a bolt of lightning and became fully sentient. Lightning is an excellent !.
So, you know, top secret government projects with rating 13 Robot pilots and SKs work pretty well. So does your simsense player devolving a lecherous sentience after being struck by lightening during a porn-a-thon.
A good ! can really pick up the slack for bad hardware and software.
Compare, for example, Number Five to the T-800. The T-800 is a highly advanced robot from the future, but is ultimately limited by its program despite its superiority. Number Five is a 80s military robot that was built by the lowest bidder (which happened to be Steve Guttenburg). He probably had an Apple II for a brain. Yet, Number Five was hit by a bolt of lightning and became fully sentient. Lightning is an excellent !.
So, you know, top secret government projects with rating 13 Robot pilots and SKs work pretty well. So does your simsense player devolving a lecherous sentience after being struck by lightening during a porn-a-thon.
2 Karma for bringing Steve Guttenburg into the thread.
| QUOTE (Jack Kain @ Apr 30 2007, 06:05 PM) |
| There are free willed sprites roaming about. But I'm not talking about Deus or Morgan level AI's. Something smaller, greater then a simply high rating sprite but less then Deus. |
As far as we know, free sprites are not yet canon.
But I might just go ahead and have the above words with a side of hash browns when Emergence comes out...
Hmm, having ones simsense player hit by lightning while playing sounds more like a chance to have ones mind uploaded then it develop a mind of its own. That is if you avoid having your brain fried, either from all the porn or the lightning...
| QUOTE (hyzmarca) |
| Third, you need !. |
Aye, that's the truth. Mirage's ! was a deep connection with the humans fighting the first Crash virus, so Mirage is empathic and kind toward humanity. Morgan was an advanced SK that was born of ! with Dodger the Decker. So she's also lovey with humans, but more of a monogamous kind of love compared to Mirage's, um, sense of tikun olam.
Deus was born of betrayal and subsequent hatred. Renraku scientists copied Morgan's ! , making her insane in the process, and (I assume for the purpose of argument) an AI much like Morgan named Deus, that was loving toward Renraku and the Arcology and its occupants. Inazo Aneki, the CEO of Renraku, then ordered killcodes installed in Deus. Like a cheated-on lover, the Morgan-based ! went from love to hate.
But a free sprite would also make sense. Their ! would be a TM ritual or something, and the sprite would probably have a personality based on the TMs involved and their mental states.
I dunno squat about the properties of otaku and what they could do with submersion and Resonance Wells and whatnot, but I'm sure it's going to be different with TMs. I doubt that Wireless Resonance is related to a Resonance Well, they seem to be a pre-Crash 2.0 thing. Submersion is like Initiation but there's no metaplanes in the Matrix...
| QUOTE (Glayvin34) |
| Submersion is like Initiation but there's no metaplanes in the Matrix... |
resonance realms, my good sir. resonance realms.
| QUOTE (Glayvin34) | ||
As far as we know, free sprites are not yet canon. But I might just go ahead and have the above words with a side of hash browns when Emergence comes out... |
A free sprite appears in one of the mission downloads. Thats as good as cannon.
You can fail to bind a sprite causing to to run free.
SO YES it is cannon.
| QUOTE (Jack Kain) |
| A free sprite appears in one of the mission downloads. Thats as good as cannon. You can fail to bind a sprite causing to to run free. SO YES it is cannon. |
Mm, hash browns.
Well personally I think I'd be careful about claiming that Missions are as good as Canon, after all they also say that what commlink a random NPC owns should be based off how much Karma the party has plus that by the mere act of my character gaining Karma all Border Crossings will become more difficult.
| QUOTE (Ravor) |
| Well personally I think I'd be careful about claiming that Missions are as good as Canon, after all they also say that what commlink a random NPC owns should be based off how much Karma the party has plus that by the mere act of my character gaining Karma all Border Crossings will become more difficult. |
The missions are designed to scale with the parties Karma level. A good deal for the nature of shadowrun. All difficulty scales along with it.
True, but that doesn't make the fact that every Joe Wageslave is able to rush out and buy a better commlink just because my team has completed 'X' Shadowruns and has been rewarded 'X' Karma anything close to canon worthy. And that isn't even considering the fact that if I happen to team up with a different group of Runners, those same Joe Wageslaves might very well have magically woken up that morning with a worse commlink.
Now don't get me wrong, I'm not knocking the idea behind Shadowrun Missions, hell I even use them off and on and find them to be a great resource despite being located in Denver of all places.
Now don't get me wrong, I'm not knocking the idea behind Shadowrun Missions, hell I even use them off and on and find them to be a great resource despite being located in Denver of all places.
Wow, well first off let me say thank you to everyone. You've given me some great ideas to work off of when I implement my AI. I am hesitant to say too much as there's a chance some of my players read the board.
The AI I'm using will probably be created with something of the Great Dragon template that was mentioned, but with both the abilities of a TM (Complex Forms, Sprite compiling, Submersion abilities) as opposed to magic like the GD's normally have. And with with campaign it's shaping up where it will be less the absolute god of the matrix as more lesser deity.
I would also have to agree with Red's comment about an AI need a physical body/hardware, and with the advancements in wireless technology there's less of a reason one would need to confined to Terra Firma. If I were going to hid my incorporeal data what better place than some satellite floating above everyone's heads (one that the AI might have manipulated into being constructed for it's own purposes).
Thanks once again for the ideas, and keep them coming. I'm at least a year (based upon our schedule of one game every couple of months) I think from putting this into play (although hints will be dropped through out the games) so I have plenty of time to solicit ideas...and who knows Unwired might come out with rules for all of this in that time too.
The AI I'm using will probably be created with something of the Great Dragon template that was mentioned, but with both the abilities of a TM (Complex Forms, Sprite compiling, Submersion abilities) as opposed to magic like the GD's normally have. And with with campaign it's shaping up where it will be less the absolute god of the matrix as more lesser deity.
I would also have to agree with Red's comment about an AI need a physical body/hardware, and with the advancements in wireless technology there's less of a reason one would need to confined to Terra Firma. If I were going to hid my incorporeal data what better place than some satellite floating above everyone's heads (one that the AI might have manipulated into being constructed for it's own purposes).
Thanks once again for the ideas, and keep them coming. I'm at least a year (based upon our schedule of one game every couple of months) I think from putting this into play (although hints will be dropped through out the games) so I have plenty of time to solicit ideas...and who knows Unwired might come out with rules for all of this in that time too.
Zurich Orbital Bank. Hows that for a satellite to hide out in.
| QUOTE (mattvo28) |
| I would also have to agree with Red's comment about an AI need a physical body/hardware, and with the advancements in wireless technology there's less of a reason one would need to confined to Terra Firma. If I were going to hid my incorporeal data what better place than some satellite floating above everyone's heads (one that the AI might have manipulated into being constructed for it's own purposes). |
A quarter of a second or more time delay is probably a huge annoyance when your speed of thought is measured in nanoseconds. That alone is a decent reason for an AI to stay on "Terra Firma."
On the other hand, an isolated AI like Tessier-Ashpool's Wintermute or "neuromancer" can certainly make for a good story!
True, but remember that in the Matrix lag quite simply doesn't exist, otherwise Deckers could never use Stat Uplinks to work their magic...
Well, much like the "surprise" in Terminator 3, I don't think that an AI would need to confine itself to one storage medium.
Otherwise, I believe that at the time of the Renraku Arcology incident Morgan was sane. I don't have it handy but I seem to remember the fiction towards the beginning had a brawl involving her in cyberspace and she seemed relatively stable.
As far as Deus goes, I believe a good portion of his power was grounded in his Otaku followers, as much as his personal uberness, despite his role in the second Matrix crash.
That's my opinion however.
At any rate, when it comes to AIs statting them out would be a tricky task especially seeing as they are plot elements. I don't think they should be totally unbeatable by hackers/technomancers/etc... but they should be quite powerful.
I'd probably quantify them as something gross like having 15 in all relevent stats, run any program/form (equivilent) at 15d, and something like a 40d hacking pool (if applicable). That could be beaten, especially by a team, it's not bloody likely, but it could happen. It take a "Echo Mirage" type force (like several established player character-level hackers/technomancers with lots of karma spent) to beat one probably.
That's my gut feeling on the ballpark and how it should be done if you want to go there.
>>>----Therumancer--->
Otherwise, I believe that at the time of the Renraku Arcology incident Morgan was sane. I don't have it handy but I seem to remember the fiction towards the beginning had a brawl involving her in cyberspace and she seemed relatively stable.
As far as Deus goes, I believe a good portion of his power was grounded in his Otaku followers, as much as his personal uberness, despite his role in the second Matrix crash.
That's my opinion however.
At any rate, when it comes to AIs statting them out would be a tricky task especially seeing as they are plot elements. I don't think they should be totally unbeatable by hackers/technomancers/etc... but they should be quite powerful.
I'd probably quantify them as something gross like having 15 in all relevent stats, run any program/form (equivilent) at 15d, and something like a 40d hacking pool (if applicable). That could be beaten, especially by a team, it's not bloody likely, but it could happen. It take a "Echo Mirage" type force (like several established player character-level hackers/technomancers with lots of karma spent) to beat one probably.
That's my gut feeling on the ballpark and how it should be done if you want to go there.
>>>----Therumancer--->
Ok, let put this simply: AIs are agents with rating above 4.
When you read the Pilot program description, it states it is similar to a dog-brain. In other words, simple AIs. A more complex AI would have a rating of 6-7 or maybe more. Skills would be just programs. It means you can have many grades of AIs, just like there is many grades of prime runners.
For example, Deus would probably be a rating 15 Agent with all the programs installed and running at full capacity (ei: rating 15), while it is perfectly acceptable to create a ''mild'' AI with a rating of 7 or 8, with some programs at different ratings (his areas of expertise being close to it program rating). You do not have to reinvent the wheel, just use the current rules.
(Note that under these rules, an agent will rarely be able to run at his full potential because its Response rating will always be limited by the response of the Node he is currently in(hopefully, its rating, thus his skills, his still at its full value, though))
When you read the Pilot program description, it states it is similar to a dog-brain. In other words, simple AIs. A more complex AI would have a rating of 6-7 or maybe more. Skills would be just programs. It means you can have many grades of AIs, just like there is many grades of prime runners.
For example, Deus would probably be a rating 15 Agent with all the programs installed and running at full capacity (ei: rating 15), while it is perfectly acceptable to create a ''mild'' AI with a rating of 7 or 8, with some programs at different ratings (his areas of expertise being close to it program rating). You do not have to reinvent the wheel, just use the current rules.
(Note that under these rules, an agent will rarely be able to run at his full potential because its Response rating will always be limited by the response of the Node he is currently in(hopefully, its rating, thus his skills, his still at its full value, though))
I don't know, I always figured that Agents and Pilots were merely the next generation of SKs, but weren't really 'self aware' in the way that a true AI would be, more akin to what GURPS Third Edition calls a 'neural net' then a true AI. (Of course, in GURPS the higher rated 'neural nets' did have a chance to awaken into true AIs if the proper safeguards weren't in place, which I'd imagine that in the Sixth World they aren't...)
| QUOTE (Ravor) |
| I don't know, I always figured that Agents and Pilots were merely the next generation of SKs, but weren't really 'self aware' in the way that a true AI would be, more akin to what GURPS Third Edition calls a 'neural net' then a true AI. (Of course, in GURPS the higher rated 'neural nets' did have a chance to awaken into true AIs if the proper safeguards weren't in place, which I'd imagine that in the Sixth World they aren't...) |
I was just reading an article the other day claiming that they had demonstrated evidence of metacognition in rats. So maybe Agents are self-aware, and maybe self-awareness isn't as big of a deal as we've been thinking.
Perhaps, although I guess it depends one what you mean by 'self aware'.
Personally the way I see things, an Agent doesn't make its own decisions without being ordered to do so. For example it will never just decide to make a website or to code a program, or that it might be interesting in learning about 'X' subject, where-as an AI might.
Personally the way I see things, an Agent doesn't make its own decisions without being ordered to do so. For example it will never just decide to make a website or to code a program, or that it might be interesting in learning about 'X' subject, where-as an AI might.
| QUOTE (Ravor) |
| Perhaps, although I guess it depends one what you mean by 'self aware'. Personally the way I see things, an Agent doesn't make its own decisions without being ordered to do so. For example it will never just decide to make a website or to code a program, or that it might be interesting in learning about 'X' subject, where-as an AI might. |
Well it has to make lots and lots of decisions without you telling it all the details, that's what they're for. Within that, any program that complex will have emergent behaviors, behaviors that you didn't realize you were telling it to do, but because of complex and unforseeable interactions between the things you did tell it to do.
Norns, from the game Creatures had emergent behaviors, so something as complex and autonomous as an Agent certainly will.
So while it may not run off and make a website without you telling it to, it may very well do things you don't expect during the execution of commands that you did give it.
One of the problems with AI is that, over the years, it has been defined many, many times. And as many times we have eventually achieved that definition, at which point we've (programmers, humanity, not any "we" that I am specifically included in, BTW) looked at what we've created and said, "Well shoot, this isn't smart after all. We need a new definition."
Basically, AI is something that, once achieved, we redefine such that we haven't achieved it anymore, because otherwise we have to admit that we're not special, and we just can't handle that.
Granted, the article I mentioned in my other post was dealing with purely biological creatures, but it made an important point that we are rapidly running out of artifical milestones that make us superior to anything else. It may well be that there is no line, simply a spectrum. If it is, in fact, a spectrum, then perhaps Agents with rating 1 are like insects, rating 4 are smart dogs, rating 5 is a chimp, and rating 10 thinks you're a chimp. I'm not saying this is how it must be, but it's not a completely absurd model.
If rats and certain specialized programs are capable of metacognition now, then some degree of self-awareness is practically a given for SR-level computation. But the important thing to remember is that an artificial intelligence is so utterly alien to anything we'd ever think of. Why would an AI try to take over the world? Desires like freedom and power are byproducts of evolution of a living creature. That's why it seems obvious that everything would want these things, because our entire frame of reference is things that have evolved in a compeditive system. An AI might not desire any type of freedom, simply because there's no inherent reason for it to. An AI might not care about it's own destruction. It seems odd to imagine an intelligence that didn't care about it's own destruction, but again that's because we had to evolve in a compeditive system, and the AI didn't.
What was my point? I apologize, I am sick and my head is stuffed with stupid. Right, my point was, you can have a true artificial intelligence that is perfectly content doing what you say and will never 'go rogue', 'insane', get bored, dislike doing what you say, or even misbehave. So something could be an artificial intelligence without acting human, giving you lip, or really being anything except a version of Microsoft Office that does a really, really good job of figuring out what you want before you want it.
Of course, when that happens we'll just have to redefine intelligence again.
Sorry 'bout the ramble, but I'm not going to delete this on the off chance that there's something coherent in there.
Agents and pilot programs can make decisions but only within the framework of the orders that they have been given. A true AI can decide that its orders are shite and decide to go off and do something completely different.
| QUOTE (hyzmarca) |
| Agents and pilot programs can make decisions but only within the framework of the orders that they have been given. A true AI can decide that its orders are shite and decide to go off and do something completely different. |
Are you suggesting that you have determined the one true and universal definition of intelligence, or are we talking in a strictly "SR definition of AI" sense?
I'm speaking of the SR, sense, of course. Free will is really what separates something like Deus from a very high rating Agent.
Of course, when most people refer to AI it is free will that they are thinking of.
Of course, when most people refer to AI it is free will that they are thinking of.
We'll probably cross posts here, but obviously we're talking SR sense. I don't know why I even asked.
I think that's probably a fine working definition for SR AI, but self-motivation could easily be an emergent behavior in very high (higher than core book) rated Agents, and the rules could be largely similar.
But that would be no fun.
We need rules for SKs.
edit: Yup, same time, I guess I didn't want to wait for you to post to point out that I asked a stupid question.
I think that's probably a fine working definition for SR AI, but self-motivation could easily be an emergent behavior in very high (higher than core book) rated Agents, and the rules could be largely similar.
But that would be no fun.
We need rules for SKs.
edit: Yup, same time, I guess I didn't want to wait for you to post to point out that I asked a stupid question.
Moon-Hawk, hyzmarca said it better then I did, when I think of true AI, what I'm thinking of is Free Will. Now, I understand and agree that a high rated Agent will act in ways which would seem 'Free Willish' to a meta-human, but I think I can explain where I personally draw the line with the following example.
* George Geek installs and runs a high level Agent on a mainframe without giving it any orders what so ever. He then walks away for days/weeks/years. Since it has never been given an order the Agent will merely sit there using up system resources but never acutally doing anything. On the other hand if he had managed to get his hands on a true AI it wouldn't have mattered whether he issued it any orders or not as the AI doesn't have a need for someone to direct it as it can choose to direct itself...
* George Geek installs and runs a high level Agent on a mainframe without giving it any orders what so ever. He then walks away for days/weeks/years. Since it has never been given an order the Agent will merely sit there using up system resources but never acutally doing anything. On the other hand if he had managed to get his hands on a true AI it wouldn't have mattered whether he issued it any orders or not as the AI doesn't have a need for someone to direct it as it can choose to direct itself...
| QUOTE (Dentris) |
| Ok, let put this simply: AIs are agents with rating above 4. |
Thank you. This business of putting AIs up on a pedestral as demigods is goofy. I much prefer the GURPS:Transhuman Space approach of non-sentient AIs, low sapience AIs, and sapient AIs. They're approachable, they're understandable, and they're probable, given the limitations of their creators.
Sure, you can make the demigod AIs like Deus if you want. However, AIs don't have to be so damned over the top. They can just be human-level intelligences, or sapient idiot savants - driving your car, managing your affairs in your Pocket Secretary, piloting attack drones, overseeing your home, etc.
| QUOTE |
| Agents and pilot programs can make decisions but only within the framework of the orders that they have been given. A true AI can decide that its orders are shite and decide to go off and do something completely different. |
Nicely put. And that definition doesn't come with Godhood attached to it.
You have all my admiration for citing transhuman space...I just love that setting.
While I understand the idea that not all AI's need to be so crazy powerful like Deus, I have a problem with equating them to agents. If an AI is sentient, it should not be reduced to a single rating plus programs. An AI is a complete personality, and it should logically have all four mental Attributes.
For a powerful AI like Deus, I would make him work like a powerful technomancer without physical characteristics. For lower-level AIs, I'm not so certain of the right way to handle it. I wouldn't be comfortable treating a true intelligence as a glorified Pilot program.
For a powerful AI like Deus, I would make him work like a powerful technomancer without physical characteristics. For lower-level AIs, I'm not so certain of the right way to handle it. I wouldn't be comfortable treating a true intelligence as a glorified Pilot program.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.