Make rogue AIs more threatening

[BaneOfGiygas]

As things stand, a rogue and murderous AI is more of a nuisance than a major threat to the station and its inhabitants. Rogue cyborgs have physical presence and can inject people full of nasty things, transport (and maybe even set up) canister bombs, deconstruct the station into oblivion with RCDs, beat people with fire extinguishers, and all sorts of incredibly nasty things.

A rogue AI, meanwhile, can set their turrets to lethal, mess with computers, remotely sap power, open, close, bolt and/or electrocute doors, and watch people. That's it. Not only are just about all of these things EASILY repairable and preventable by a competent mechanic or even competent people at all, but cyborgs can do all of these things with just as much ease as an AI. The one thing, the one single thing that the AI has over the cyborgs is being able to see through the security cameras, and even THAT'S preventable by the savvy crew member. Furthermore, AIs have zero mobility whatsoever, which can be a rather large problem.

Basically, rogue AIs need to be more of a threat. They need to have more capacity for threat, they need to be able to do things aside from bolting and electrocuting all of the airlocks and setting everyone to arrest in the security records. Everything the AI can do can be reversible in mere minutes, and while I'm not saying that the AI should be completely unstoppable, it can't be completely useless either.

One suggestion I had in mind was to make the AI shells agents of further AI destruction. Give them some fancy toys to potentially wreak havoc with (or do helpful things with, if they're not rogue) or let them hijack robots/cyborgs, maybe let them remotely cause robots to malfunction, something, I dunno. In any case, rogue AIs always feel like a second wheel to the death squad of rogue cyborgs; they're supposed to be the big bad overseer when they go rogue, so let's help make them big bad overseers instead of that damn robot that keeps bolting the doors.

[TailorTok]

I agree, I think shells having lasers would be nice. Shells are super weak.

[Frank_Stein]

Something I really want to see is the AI having some kind of control over the normal bots that can be build. Medbots, firebots, floorbots, securitons, those things. Like, just the ability to tell them to move to a certain area would be enough.

A helpful AI could respond to station crisis and generally be more helpful that way. When the AI has been corrupted and is working with a human, it could send emagged bots out to mess up areas. Barring that, it could just manually add people to the arrest list and send Securitrons out to hassle em.

[BaneOfGiygas]

Something I really want to see is the AI having some kind of control over the normal bots that can be build. Medbots, firebots, floorbots, securitons, those things. Like, just the ability to tell them to move to a certain area would be enough.

I know that AIs can remotely turn them on/off, but that isn't very much...and, again, cyborgs can do it too.

[Grizzwold]

Barring that, it could just manually add people to the arrest list and send Securitrons out to hassle em.

You can already sort of do this with the securitron program on your PDA. The summon bot command brings the securitron to the currently selected camera. Of course, then you have to deal with the godawful PDA interface and associated lag.

[BaneOfGiygas]

Barring that, it could just manually add people to the arrest list and send Securitrons out to hassle em.

You can already sort of do this with the securitron program on your PDA. The summon bot command brings the securitron to the currently selected camera. Of course, then you have to deal with the godawful PDA interface and associated lag.

Not to mention the fact that this is easily foiled by any logged-in schmuck at a security terminal.

[Lantern]

stuff.

I think that rogue AIs are fine. The most deadly rogue AI is the subtle ones.

I will admit however, that being a rogue AI can be more than difficult when there are no borgs around. But still, as it is, you can kill plenty of people as a rogue AI as it is. You just have to know how.

[Erev]

AIs need more to do in general. Sure, you've got a couple options with human help. And a few more if you know the computer systems well enough (few do). Still, it should be more hands on and entertaining than it is. And this definitely includes making it more dangerous when corrupted.

[BaneOfGiygas]

I admit however, that being a rogue AI can be more than difficult when there are no borgs around. But still, as it is, you can kill plenty of people as a rogue AI as it is. You just have to know how.

The only bodycount a rogue AI is really able to rack up is through crew incompetence, (e.g. repeatedly running into electrocuted airlocks or into lethal turrets) cooperation with other antagonists, (a rare sight, as most antagonists are content to upload a rogue law, grab popcorn, and kick back letting the AI do its thing) or lightning fast reflexes with the door controls. (Incredibly difficult when you factor in lag and the less-than-optimal controls)

The overall problem is that the AI has no ability to effectively INITIATE danger. Airlocks and APCs can be hacked, turrets can be destroyed, shells are almost useless, and security cameras can be snipped, all of which forces the AI to hope that no one's competent enough to grab hacking tools. One crew member with electrical gloves and a blue toolbox, one single competent crew member can irrevocably cripple a rogue AI and render everything they can do absolutely useeless. The AI has to wait for crew members to bring themselves into danger, which is really difficult when you consisider how quickly rogue AIs get called out.

Hell, the fact that a pair of fucking gloves is all someone needs to be pretty much invincible to the AI is rather fishy in and of itself.

[Zafhset]

I'm going to differ on this. Have you seen someone like Herr Spy Guy play a rogue AI? It's terrifying.

An all seeing entity with decent knowledge of telesci and the ability to track your location is like having a sniper paint a crosshair on you. That's not even all there is to what the AI can do to cause harm.
It's not that rogue AIs are not threatening - it just takes a certain amount of know how. Shells are already pretty robust as is with their immunity to flashes and conventional stuns.

[icarus]

turning the safety mechanisms off on doors so they crunch people who walk through them could be a pretty neat rogue addition

telesci is an underrated tool for rogue AIs, but if anyone touches the APC, it's immediately not an option. this applies to most rooms, too

[BaneOfGiygas]

An all seeing entity with decent knowledge of telesci and the ability to track your location is like having a sniper paint a crosshair on you.

I have a very solid idea on how to do telesci, but this is more complicated than you seem to give it credit for. It requires having a scientist who's willing to move the GPSes around for you and share the readouts, which I've never in my entire career as an AI been able to arrange. Not only that, but given the rather notoriously iffy camera-switching mechanisms and the lightning-fast calculations you need to be able to make, trying to efficiently use telesci as an AI, rogue or otherwise, is really damn hard unless you have a helper of some sort.

Shells are already pretty robust as is with their immunity to flashes and conventional stuns.

Immunity to stuns does nothing to help their cripplingly fragile nature, lack of effective combat tools, (the most damaging thing they can use is probably a crowbar) and perplexing lack of ability to recharge as far as I'm aware.

Have you seen someone like Herr Spy Guy play a rogue AI? It's terrifying.

I have not, but I'm willing to rule this example out as an exception because he's goddamn Spy Guy. That man is on another level of terror.

[Grayshift]

You don't need GPSes to work out the telesci values. It takes a little longer the other way, but it's worth it.

I should play AI more often. CHORUS will continue in Tranquility's footsteps.

[Frank_Stein]

AI and Robotics are definitely due for some tweaking. Both suffer from a lack of things to do past crisis situations, and in the case of AI there's just some real UI problems.

I'd like to see a risk vs. reward aspect to the AI and the crew. Like, over the course of a round the crew can upgrade the AI to be more useful, but also more dangerous if subverted.

[Zafhset]

The crowbar does suck as a weapon. Any two assistants can out beat you with an extinguisher. Shells are best used in none direct confrontations - push folks into space or into electrified airlocks (the gloveless ones), become a pod flying menace. Bump welding tanks into light sources.

It's true that it's fairly easy to take away some of the capabilities of a rogue AI, but that's where cooperating with your fellow borgs and the folks that uploaded laws come into play.
Borgs can fix cameras and restore APCs as long as it isn't damaged or had its battery chucked out. your human conspirator (if any) could do anything with the right tools and equipment. You complement them by being able to track and check things for them, and by having a measure of control in the environment they travel in. They complement you by keeping your capabilities functional (sight, telesci, etc) and securing your core when it gets raided by tooled up individuals.

No borgs? Either convince the human to borg some people or cripple genetics and wait until you have a coterie of borgs before going all out.

Malfunction isn't an actual game mode because of the relative ease in which you can upload a law or reset the AI.
Making the AI any more capable than it is right now would just completely devastate an unprepared crew. Antags shouldn't just be able to turn a blind eye on core security after subverting the AI.