Base AI

[S.T.A.L.K.E.R.]

So I was thinking, right now some of the energy use is really inefficient. If I want to preserve energy but also keep the place well lit, I need to manually turn lights on and off whenever people enter or leave the room. If I want turrets to save energy, I have to go into the menu and click them off myself, do I want the nutrient dispenser to stop soaking up those watts? I have to click it off then on when someone is hungry.

I was thinking, maybe a late game research project could be a base AI. The AI could turn lights on and off dynamically when people are entering or leaving the room, as well as change turrets for the location of hostiles and the other electricity soaking objects. Mostly akin to a power switch. The AI would be controlled from an AI Core building, a large sphere and a console for the colonists to configure it.

The AI could potentially go nutters and try screwing with your people, by doing stuff like turning turrets against you and locking people out of rooms. Of course if it does that, the only way to stop it is to get to the AI core and shut the bastard off, but it'll try to stop you.

[FtDLulz]

Sounds like a pretty interesting idea. I like it. I assume the AI would take up quite a bit of power though.

[S.T.A.L.K.E.R.]

Sounds like a pretty interesting idea. I like it. I assume the AI would take up quite a bit of power though.

Yeah, it'd probably be difficult to run it with just solar. Speaking of that... *begins plotting new threads.*

[StorymasterQ]

"Base AI, open the door!"

"I'm afraid I can't do that, Dave."

[Darkfirephoenix]

Why not let it help with other stuff? Like research, or controlling robots (if they get implemented) and maybe automated Stock-keeping. So if you say that you need at least 400 raw-food of any kind it will try to stock at least that much or you could tell the AI to produce some weapons wich normally would take a ton of time for an Human. It would really well work with other stuff we could put in: Auto-Medlabs, Auto-Hydroponics etc. It would take a ton of work from your ppl so they could do other/more important stuff, like looking for an way off this Planet / a way to Terraform it...

[S.T.A.L.K.E.R.]

Why not let it help with other stuff? Like research, or controlling robots (if they get implemented) and maybe automated Stock-keeping. So if you say that you need at least 400 raw-food of any kind it will try to stock at least that much or you could tell the AI to produce some weapons wich normally would take a ton of time for an Human. It would really well work with other stuff we could put in: Auto-Medlabs, Auto-Hydroponics etc. It would take a ton of work from your ppl so they could do other/more important stuff, like looking for an way off this Planet / a way to Terraform it...

Having the robots work through the AI systems so they go bonkers when the AI does is neat. Maybe sufficiently smart colonists can rewire an insane bot and make him work normally again.

[Darkfirephoenix]

Maybe even make a AI needed for some sorts of power-generation, or at least if you want no accidents (Fusion-Power is pretty complex to keep in check, but the worst wich could happen is that the Reactors "just" melts and sets things on fire [not as dangerous as an Fission-Reactor], or Power from Anti-Matter). Maybe the AI is needed for some Research too (or you would need a ton of calculating power/time).

But the way to an AI should be more like this: .... Self-Learning Programms, Quantum Computing (or something of that kind).... AI
Just being able to research and instantly building an AI would be... well, a BIT unrealistic (unless you get some stuff to kick-start it).

If you make an AI from scratch: You would first need someone wich knows enough about programming to write the "rules" for the AI so that it is "tame" (or you don't do it/cut the time a bit, but it should increase the risk of the AI going crazy).

If you make the AI from an Intelligent-Programm: There are the main infos already existing wich would cut the risk and the time needed to programm the AI, but an person with enough knowledge is still required to ensure that the AI is programmed correctly.

And finally: You find an AI! .... Are you sure that you want to entrust your life to an rabdom AI, wich dropped onto the planet not too long ago (or you found it in a ruin/wreckage).... At least you should check how it's programmed and wich rules it has got... (this would cut the needed time REALLY, but even if you have an person with enough knowledge there could be still some "hidden" parts of the programming [wich would increase the risk of the AI going crazy. Who said that the AI wasn't THE CAUSE/REASON of the accidents... ])