How do you Hack a system that has an AI?

[AlanGunhouse]

An AI is an intelligence, you do not hack it, you convince it. Run it as a NPC and have the hero talk it into helping.

PS, you might be able to cyberjack specific systems, like a door or a single weapon, but not the whole AI.

[AlanGunhouse]

As I noted in my edit while you were answering: you can probably cyberjack specific systems...like a single door, security cam, or weapon.

You could also attack it...but I am not sure how you would handle that part.

[Nightmask]

As I noted in my edit while you were answering: you can probably cyberjack specific systems...like a single door, security cam, or weapon.

You could also attack it...but I am not sure how you would handle that part.

One of the HU2 books had detailed rules for hacking I believe, including the impact powers like Mechano-Link have on things. If you were trying to hack an AI it could be taken as a form of virtual combat with the PC's avatar battling the AI avatar in its own environment as you attempt to subjugate it. While you can't run roughshod over it like a 'dumb' system it shouldn't be impervious to cyber-combat and likely defeat (Mechano-Link gives massive bonuses) either.

[Nightmask]

As I noted in my edit while you were answering: you can probably cyberjack specific systems...like a single door, security cam, or weapon.

You could also attack it...but I am not sure how you would handle that part.

One of the HU2 books had detailed rules for hacking I believe, including the impact powers like Mechano-Link have on things. If you were trying to hack an AI it could be taken as a form of virtual combat with the PC's avatar battling the AI avatar in its own environment as you attempt to subjugate it. While you can't run roughshod over it like a 'dumb' system it shouldn't be impervious to cyber-combat and likely defeat (Mechano-Link gives massive bonuses) either.

Yes. I know about that section on Cyberjacking/cyber-combat in AUGG. But I don't think it applies as my hacker is not in a cyberspace environment. And how do you determine the SDC or Hitpoints of an AI?

Really depends on how advanced the AI is and how much anti-hacking features it has. There's a lot of play in that regard, and much like someone hacking a Cyberjack's brain in Rifts. The AI is effectively that brain and the built-in defenses are its AR, SDC, and possible HP. A cutting edge or the most advanced around is going to be far better defended than a run of the mill AI especially if its builders didn't fear much risk of hacking (say it was a closed system without outside communications and personal security was pretty tight to keep out an intruder) then it might have only rudimentary firewalls against hacking.

[NMI]

It would probably be best if you did introduce the cyberspace environment.
If I recall [books not in front of me], the HP and SDC of an AI or any artificial entity in cyberspace is determined by there IQ and any filter [armor] programs they have.
If you throw away cyberspace [gasp!] then you throw away the need for HP/SDC in cyberspace. You have basically brought this down to
a: roleplaying interaction with the AI
and
b: a couple of die rolls based on the penalties under the Power Category: Hardware: Electrical.

[Nightmask]

Hmmm... Good point. Although I'm still confused on how I would run it in the event that my players made the AI hostile, and a fight ensued... I will reread the section on Cyberjacking again.

Maybe I'll just roll up the AI like an NPC.

Question, how would you run it? Need an example if you don't mind... My brain not really working right now...

Haven't got the right books handy to really think of a good example, although I think under Mechano-Link didn't it require the AI to be at the level of self-aware/sentient? If so it'd be a moot point as it'd be as vulnerable to complete override as any other computer. If he were to end up in cyber-combat with it probably not a bad idea to just make the AI with an avatar giving it the basic stats useful in virtual combat, determine what kind of avatar the PC has and let him battle it out. If he wins the AI is overridden and subjected to his control, if he somehow loses he gets ejected from the mental strain of cyber-death and takes some temporary HP damage (since it'd be damage inflicted on his brain causing real physical symptoms) he'll have to heal.

[LeeNapier]

How do you Hack a system that has an AI?

With an axe.

It sounds like a joke, but it's not - that's probably what I'd do. Just locate the AI core and physically destroy it.

[Shadowfyr]

I actually have someone in my group with Machine Merge and has wanted to try this scenario out to. I just created a cyber world (similar to the Psionic power Machine Ghost) created what the facility would look like dependent on how powerful the A.I. is , the Player retains their sdc/hp and digital similar weapon. and the A.I. as NMI said about its IQ level plus what other "armor/Defense" programs for its HP/SDC. and just create the awesome digital battle.

And if only the one character has the power let them be in the Cyberspace and make everyone else hold down the fort/protect the character's body/ or even try to simultaneously hack in other programs to assist the PC in the Cyber world..

Just some ideas.

[Mercdog]

Just spitballing here, but I think I'd just have the player spend some time making Computer Hacking rolls, with a penalty based on the AI's capabilities (I don't know, say a base minus 10-40% depending on the AI). For each success, he can order the AI around for a melee round. Or you could set number of successes needed, like best 3 of 5, for the hacker to completely bypass the AI's defenses in order to get the information your team wants.

[Nightmask]

Looking around I just happened to notice that the Aliens Unlimited Galaxy Guide actually explicitly notes what effects Mechano-Link has on hacking, along with a number of other powers (including spells and Nightbane talents) that have an impact on hacking computers including AI.

[Sir_Spirit]

Yes. I know about that section on Cyberjacking/cyber-combat in AUGG. But I don't think it applies

It does.
If you have hackers you have cyberspace, regardles of how primitive it is.
Be it phone lines, the "interwebs" or the Matrix.

[The Beast]

What if it becomes hostile?

Then it'll probably send something back in time to kill your mother, and if that fails, it'll send something else back in time to kill you as a kid. It however will not send anything back in time to try to kill your great grandparents when they are kids. It'll just keep sending things to kill you, just at a later date than the previous attempt.

[AlanGunhouse]

What if it becomes hostile?

Then it'll probably send something back in time to kill your mother, and if that fails, it'll send something else back in time to kill you as a kid. It however will not send anything back in time to try to kill your great grandparents when they are kids. It'll just keep sending things to kill you, just at a later date than the previous attempt.

That is because sending things back to a date before the first date it sent things back to is prevented by the way the time machine works

[BillionSix]

Reminds me of this:

http://www.applegeeks.com/comics/viewco ... ?issue=289

[Snowtiger]

Well, on a pure technological level hacking a computer with an AI running on it doesn't necessarily differ that much from hacking a computer with non-sentient programming running on it. The AI is just another program running on a computer, albeit in this case, it's a mighty complex program, and acts more like a permanent system operator rather than just a program running it's course and then being freed from memory when done. It may have devised a series of very tough firewalls for protection and usually will take offense if these defensive barriers are broken into, just like any human system operator would take offense on an intruding hacker in their system and react accordingly.

As the AI is more like a system operator than a common program, it should have all the stats necessary to appear as a system operator. This means that it probably would have all mental stats (IQ, ME, MA) and have skills related to what it can do. It basically knows how to operate sensors and other computer-controlled equipment (for an example an AI supervising a computerized manufacturing plant would be able to control assembly lines, CNC machinery and other things like that, and be at least knowledgeable on how to use a CAD program or two).

The important thing with AIs is that there are very few general purpose AIs, that can freely do anything they wish. The programmers of the AI usually create it with a specific task in mind, so it well be good in the chosen area of expertise, while tasks outside the defined area of expertise are more likely to fail. If an AI hasn't been programmed to do something, it needs to be programmed to do that task, or in some cases, needs to adapt its programming to do the task (adaptive programming in common AI's is scarce, but an advanced AI(such as the one in the Robotics section of HU2) utilizes adaptive programming to learn how to do things on it's own or with a suitable tutor).