The basic question i have is meat creatures can upgrade how their minds interface not only with their bodies but also the matrix and the rest of the world. Would it be possible to buy things like and encephylon or an attention co-processor to help the A.I.? If these upgrades could be made to a particular comlink its using, how would you go about costing them, or installing them, and if not what would be preventing the A.I. from benefiting from these systems?

Unfortunately you can't get something like an encephylon for an AI because they lack the physical body on which to put it. This is one of the many many many reasons that AIs are considered underpowered in general. They lack a physical brain, which is what all hacking type upgrades to a person require as an interface.

It doesn't make much sense that a good hacker can have 5 IP, but a computer program (IC AI Agent) only has 3 IP(Or is it 4?), somehow the metahuman's brain is able to operate faster in cyberspace than cyberspace itself can operate, but that is just how the rules are.

For an idea of a fix, allow technomancer initiation for AI's. Swap out Karma for Karmax1000 Nuyen.
Or Karma and Karmax1000 Nuxen.
Or allow things like the encephalon to be built into the Hardware of your Home-Node.
Do you trust someone to do something like that to basically your home/body/soul?
Sooner or Later, you would have an AI with enough Upgrades living in, basically, a
Supercomputer/Server-Rack or something. Which would be fitting, neh?

Well, I don't see why an AI would be any less likely to trust someone with their home node than a human would be to trust a street doc with their body. Don't forget they already have to have people come repair and run maintenance on their home nodes (As evident by having to pay lifestyles).

As for the encephalon, there are two big problems. First off is the fact that encephalon technology is specifically designed to work on a brain. An AI doesn't have a brain so the equipment has nothing to operate off of. The other problem is that the AI moves about in cyberspace, it doesn't always inhabit its home node, and so would usually not benefit from any sort of attachments to its home node.

As I said earlier, this is among the many reasons that AIs are considered horridly underpowered.

Clone a body, implant comlink, use as homenode, implant encephalon?
Would explain upkeep cost as well.
Do those home node things really not work when outside of that thing?

Still doesn't work because the AI usually isn't in the home node. AIs can become trapped in a nod if it suddenly becomes disconnected from the matrix while they were accessing it because they actually go into that node instead of just connecting to it from afar like everyone else does.

Also, to my knowledge, you can't use a human brain (Technomancers aside) as a node, so having a clone body around wouldn't help the AI any. Just because there is an encephalon in the AIs general vicinity doesn't mean that it can make use of it.

No, cloned brain, implanted comlink, make that comlink your homenode, add in encephalon, use brain solely as connecting tissue, nothing more O.o

Yeah, but the encephalon works by interpreting the signals in the brain, not just using it as connective tissue. So unless the AI was actually inside the brain, and using the brain as the home node, it wouldn't make any difference. And at that point you may as well just make a cloned character and say it has was inhabited by an AI (Which I think there is/was a thread of somewhere)

Remember, bio/cyberware affects the person, not the commlink.

I've had a few questions on this as well, if the AI is using a commlink as a home node and the commlink is modified does the AI get the benifit? IE Extra IP passes?

What about raising attribute scores over 6? possible?

No and no.

The thing people seem to have trouble with is that an AI doesn't use a home node like a metahuman uses their personal commlink. Instead the AI uses their home node like a metahuman uses... well a home. They live there, they heal there, they hide there, but they don't do their shadowruns in there. When they are hacking a corporate node, they are cyberphysically moving into that node, they no longer are in their home node any more than a metahuman runner is in his home while he is in the middle of a corp building. Thus an AI would gain no benefits from his home node, even if the things that you can put on a commlink could help the AI. Remember, virtually every hacking related upgrade is focused on improving the brain, or improving the connection between brain and commlink. Since the AI has neither, it can benefit from neither.

Hence the reason that I really like the idea of allowing AIs to use the Technomancer upgrades...

I know this would be a houserule but why not allow the various cyberware upgrades that would be appropriate for an AI but as application patches/updates/additions to the AI's source code?

Thus an Encephalon becomes a piece of software that performs the same tasks / gives the same benefits as the hardware does in a metahuman. It's not like in 2072 they cannot emulate in software anything that hardware can do.

Thus you could have standard / alpha / beta / delta software just like regular cyberware. Increased costs based on grade, etc. That would mean Delta grade AI programs would be custom made for the AI like the actual hardware is for the implantee.

Maybe even have an 'Essence' rating equal to the AI's Rating to cap just how much can be changed of the source code before it starts to fall apart.

That actually sounds very reasonable, and I think I like the idea a fair bit.

That actually sounds very reasonable, and I think I like the idea a fair bit.

Poor Tux... he must be in shambles by now.

Or maybe he has delta grade software upgrades

That's a fantastic idea. Gonna have to run that by our GM and see what he thinks.

Thinking about this idea of the AI not having a brain just made me think of the very end video of System Shock 2, where

*Copy* and *Paste*. Hehe. Nice. I've been having issues with the AI inferiority for a while now. You may have just solved it at my table. Have to discuss it. Definately a start. Must go experiment now.

Glad I could help all.

Or if you don't want to be inferior....don't play an AI?

Because AIs shouldn't suck in the 'Trix, period.

I'm not sure why that's the thought, your saying that something the evolved from a toaster in some cases should be equal to a metahuman who's trained and specialized and prepared for their chosen role. Furthermore just because all choices are valid does not mean all choices have to equal.

But I admittedly have a bias against the mini-AI's.

No, but I am saying that something that has evolved in the shifting data of the 'Trix should at least be on par with a metahuman who at best has to have her thoughts relayed through mutliple layers of tech and translated before they are able to do anything in the 'Trix.


But then again, I'm a fan of the new "mini-AIs" the old godlike ones were just too much like Great Dragons meet IEs. I did like the Arc Shutdown though, but I think a properly run mini-AI could have pulled it off as well...

A gold fish that evolves in the shifting waters of the pond can still be eaten by a frog.

The old AI's while "big" concerns did have a certain mystique about them. The current "Oh hai I'm an AI I evolved in your eurocar." just seems stupid and a further effort to dumb down the matrix side of the house.

Furthermore human brains over tech is a core conciet of the setting, otherwise there'd be no echno mirage no matrix none of it. So saying that a being that "evolved" (and I hate that term by the way) in the matrix should somehow be better would overturn a lot of history.

I think it'd be interesting to have some upgrades for AI's that you could buy and install in their home node. It'd be kind of like cyberware for the AI character. Maybe that'd be something good for someone to submit to the Dumpshock Data Haven publication.

Sure I see that point, but in a fight between "equals" I'll put my money on whichever one is on their "home turf" so to speak, and I haven't seen anything that says a "human level" AI is somehow inferior to humans and in my opinion the fluff disagrees with the rules, it is supposed to be hard and dangerous to go "AI Hunting".


the_real_elwood the only problem with that is that any upgrades to the AI's home node is only going to be useful while the AI is there, which is going to be almost never.

Honestly, I don't even think that the biggest weakness of an AI is the fact that it can't get all the amazing gear that a metahuman hacker can. I think it is the fact that they must put themselves so directly in the line of fire to do what a hacker does with only a minimal amount of risk. If a node gets turned off while a hacker is in it, he suffers some dumpshock, goes 'ouch' and starts doing other stuff. If a node gets turned off while an AI is in it, it is effectively dead until the node gets turned back on and reconnected to the matrix.

In other words, AI hunting is really really easy, you just get them to come into your commlink or some node you have physical control over, and then turn it off and scrap the thing. I remember how there are some lifestyle disadvantages along the lines of 'an AI has taken up residence in your home node.' and I can't help but think "Okay, I turn it off and buy a replacement, problem solved for a handful of nuyen."

True, but that doesn't really brother me since it makes sense that a critter living in the 'Trix would have the same dangers that we would when fighting something from the fourth dimension, besides, it probably isn't easy to trick an AI into a Node that you can simply pull the plug on.

Well, they can't -do- much to you without doing so, so it might not be all that difficult.

Touche'

I'm just currious what keeps and A.I. from having 2 matix windows always open at once, 1 to the target node and one to a safe node with a massive firewall and the like, so that if the target node is ever shut down he gets kicked into the safe node by default?

I'm just currious what keeps and A.I. from having 2 matix windows always open at once, 1 to the target node and one to a safe node with a massive firewall and the like, so that if the target node is ever shut down he gets kicked into the safe node by default?

The fact that moving around on the matrix isn't like using firefox. A persona must actually enter a node in order to access anything from that node or do anything on the node. For metahumans that persona is a program that is generated through their commlink (And can be regenerated easily if it gets trapped), for an AI, the AI itself is the persona that must go from node to node, and it can't be backed up because it is too complex.

The reason it can't just hop to another node when the current node is either turned off or disconnected, is that the only way to know that the node is turned off or disconnected is for it to actually be so, by which point it is too late (No magic my 'someone's hand is near the power cord' senses to warn an AI that the power to the node is about to go)

I believe there are actually rules for having access to 2 nodes at once, unless that was an optional rule posted here at some point.

I'm guessing you mean the 'connections' thing. Having a connection to a node and actually being in a node are two very different things. A connection is like having IM capabilities while actually being in a node is like being in the house of the person you are IMing. Sure you can talk to the person via IM, but you can't affect their house in any way.

SR4 pg 218 would seem to differ with you. At least in my reading of it. It seems to indicate you can engage in cyber combat in multiple nodes and be attacked in those multiple nodes simultaneously. This implies you do not have to be present in a node to effect or be effected by it. So why couldn't an AI simply stay in one relatively safe comlink carried on or in another party member, and reach out from it to do his thing in the world at large.

Because a commlink can generate multiple personas, an AI is a persona and cannot copy itself.

try page 110 of unwired, and the text on node movement.

i think the end problem is that AI's are not really fleshed out, as they are a new element of SR (the older ones where more like boo scary gods with no stats, only gm fiat).

there are also some qualities hiding in unwired, and priced in a runner companion sidebar (i am tempted to say that one should be able to buy those as the character evolves, not just at generation, as the AI is code, not flesh).

Thanks, I was pretty sure I'd read this somewhere before, I just couldn't find the reference. So an A.I. can be carried in a runners com link and do their hacking on the fly then.

yep, it may even make said comlink its home node.

However, bare in mind that Unwired needs some serious errata in order to bring it into line with Fourth Edition 2.0 so the "advanced rules" needs some tweaking.

yea, i think there is a contradiction between agent movement in SR4A and unwired, tho personally i vote for the unwired way, even if it brings some level of asshattery in terms of agent abuse. At least that way the AI's are not limited compared to meat users in terms of connectivity...

tho i guess they could say that the SR4A limit is only for a properly limited agent, while AI's and unlimited agents behave like unwired...

Can't an AI only be killed by attacking it in its home node? Wouldn't a smart AI have a wicked firewall'd home node so it can go out and engage in matrix activities without being put in serious danger? The AI program's code is physically installed on a machine that actually exists somewhere. How can you permanently damage it by attacking its manifestation on another machine elsewhere?

well you can lure its running copy into a node, then isolate said node, you can kill it.

see page 90, runners companion, last section of realignment and restoration.

Complete immunity to biofeedback is pretty exceptional, but that's not the matter at hand. The matter here is that you're all thinking incorrectly on how an AI should operate. A straight off chargen AI with a system of 6 can have active subscriptions to 12 agents running around at any given time. That's 12 decoys and 12 hands. What are the chances that the node someone set up to trap the AI (free birdseed for roadrunners?) actually has the AI in it? An AI should shun such common and disposable nodes as someone might casually destroy and only hisself enter nodes of note and worth.

Besides most corps would rather convince an AI to take up residence in their systems than destroy them. An AI doesn't have the same kinds of concerns (read: politics) a living creature does, so it would be easy enough to satisfy one, and while it's held up in their system, that selfsame system is running at SotA levels. We're talking system or response ratings of 8-10 for an already powerful machine. What profit-driven org wouldn't love to recruit an AI?

Yeah, sure, an AI can have 12 agents running around, but so can any metahuman with a decent commlink. There really isn't much of anything that an AI can do that a Metahuman can't do much better.

Which is the primary reason I'm wondering about an A.I.'s ability to obtain upgrades to its programing.

Decent comlink? You mean military a grade supercomputer stuffed into their head? Yeah. That's true, but it wasn't my point. My point was that it's harder to pin down an AI than "tricking it into a comlink and smashing it" since who ever was monitoring the trap shouldn't really be able to tell the AI from its agents. Besides, all you have to do to kill a street sam is trick him into a building and then blow it up. Doesn't mean that's good storytelling.




Yeah, including die. Complete biofeedback immunity means that an AI can rip some of the toughest of IC to shreds and can never get trapped into a combat. Decent benefit imo.

Besides the software emulation of implants, here's another option for GMs who want their AIs to be less limited - AIs aren't natural, therefore they shouldn't have natural maximums, just starting maximums. Give them the room for unlimited, slow growth as they mature. I mean the basic idea of these AIs is evolving code which happened to evolve self-awareness and self-determination. They should have the potential to become hyper-specialized.

And besides all the rules and suggested rules, AIs are just fun for roleplaying. I'd be happy if they replaced hackers entirely. Pasty white nerds roleplaying pasty white nerds is .