The AI and its future on Aurora.

[Pratepresidenten]

As its been over two months since the whitelist was implemented, we're confident that this has been a step in the right direction. So with that, Im making this thread in an attempt to better the AI further.

We're looking for suggestions and ideas about how to improve things in regards to the AI. So far there has been ideas thrown around about giving the AI a type of mobile shell for a more physical presence. To remove/rebalance cameras and how the AI is integrated into the station. And lastly to move the AI core somewhere else. Perhaps more centralized and/or less fortified.

 

So with that, we'd love to hear your ideas and suggestions. No idea is too big or small!

[Colfer]

I have a few suggestions and ideas;

1. is to make cameras blink very noticeably if the AI (or anybody on the camera monitor) is looking through them, perhaps blue if its the AI, and red if its security. It would be a good idea to make all the cameras within a large (bigger than the AI's vision) area (such as 16 tiles) have this affect, so that funky logic doesnt make this feature worthless

2. allow the AI some method of directly operating the station to some degree, mobile shells are a must have when you as the AI know medical to an M but theres no medical staff and you just have to watch people bleed out from arterial ruptures. I think cyborg shells would be best, but IPC remote shells would be good as well. Mechs are very situational but I feel they would be best accepted by the community.

3. Update the description of the agent ID card to inform the buyer that it stops the AI from being able to see you entirely when electrowarfare is on (unless it already has this description) OR change electrowarfare to be "Stationbound Cloak" or some other straightforward name.

4. I heard a suggestion about making the AI have multiple cores it can transfer to and use at a given time, I think this is a good idea but it would take alot of work to be implemented, to further expand on that idea, if it was potentially added I would think that each department has 2 cores, the AI transfers into the cores to have control over that department where destroyed cameras will just enable a mesons like view instead of totally isolating the area from control. The AI could still have a centralized core and simply transfer controls to specific areas and is able to be intelicarded from those areas if it chooses so, but being intelicarded at your core is not a choice. Destroying both cores in a department would completely isolate that department from the AI's control regardless if it can or cant see the area

5. Add a hacked AI intellicard to the traitor uplink thats quite expensive (10-20TC), this comes with remote download capabilites and allows a traitor to have an AI in its pocket and potentially manufacture one if they have the right tools. The intellicard would have wireless control and radio functions off by default to avoid any accidents when the AI joins on the card (but able to be turned on by the user). This hacked AI, when inserted into a core would have a custom make-shift syndicate sprite (unless it has a custom sprite in which its overridden) and cant be changed and would also drain less power (8000w maybe, basically the minimum to stop the AI from instantly dying when its put into a core because no area supports its level of power drain except the AI core). The AI would start with the syndicate laws and not be connected to an AI upload as well as having traitor communications and coming with a traitor radio key for the buyer. (Optionally, also restricting it from having every radio channel, meaning you still require to put it next to an intercom if you want it to intercept communications. This may make it less convenient for the antagonists given the setup required though.). This is so Antagonists don't feel like the AI's power is ever out of reach no matter the security in place. It would still very much require a whitelist however

Edited May 10, 2021 by Colfer

[thebest_player757]

2 hours ago, Colfer said:

1. is to make cameras blink very noticeably if the AI (or anybody on the camera monitor) is looking through them, perhaps blue if its the AI, and red if its security. It would be a good idea to make all the cameras within a large (bigger than the AI's vision) area (such as 16 tiles) have this affect, so that funky logic doesnt make this feature worthless

2. allow the AI some method of directly operating the station to some degree, mobile shells are a must have when you as the AI know medical to an M but theres no medical staff and you just have to watch people bleed out from arterial ruptures. I think cyborg shells would be best, but IPC remote shells would be good as well. Mechs are very situational but I feel they would be best accepted by the community.

I agree with most of these, specifically number 1 and 2. Part of the AI's power is simply not knowing when it's deciding to look at you, and this leads to the worst of the AI hijinks I think, when an antag is just minding their own business and suddenly their entire gimmick tumbles due to a situation out of their control. I believe on baystation the AI could operate maintenance drones, which were capable of doing some rather basic station repair and whatnot, but having a way for them to do simple things at the cost of giving up their omnipotence for a moment seems fine enough.

[Butterrobber202]

Speaking of omnipotence.

I think it would be neat if we used those spare core slots in the AI core for other mini-AI's. You know how changelings absorb players into them and they can communicate and stuff? Maybe give the AI a similar ability where ghosts can hop in as these subservient cores and help out a busy AI. They'd only be able to talk to the Master AI, but they have all the controls of an AI. Additionally, I don't think this role would be whitelisted, because it would provide a good taste test to being the AI. (And because the AI should be able to eject Sub-AI's that are not acting right.) I think it could push AIs in a more fun direction.