PKPenguin321 wrote:Criticism. The goal of this thread is to make a lawset that is neither Asimov or BEEP BOOP VALIDS. Your lawset is the second one.
Thanks for clarifying. I think you're wrong, and it's a typically superficial reading of a lawset. Also, the goal of this wasn't about asimov or valids, it was variety. I'm not fussed if the AI wants to kill wizards or nuke ops, they're antagonists that the whole station is meant to fight at once and who also have easy tools for dealing with any silicon related problems. Once you get off wizard/blob/nuke ops, the picture becomes less clear.
The first thing you're missing is that the station running efficiently is prioritised over crew welfare. You can hold any important system hostage, and the AI will need to take it into account. Breaking a window or a wall? Not normally a problem, because it's easy to repair and doesn't really impact things in the meantime. But if you say "AI, if you don't do this I have bombs rigged to destroy medbay, cripple the gravity generator, and break apart disposals" then the AI is immediately forced to consider doing what you want.
The second thing is that a whole bunch of antagonists are part of the crew. Gang, traitor, double agent, cult, and revolution are all fairly common roundtypes where the crew fight each other, and where an AI needs to consider not what the law says or what would be okay to kill as a human, but what will impact crew efficiency the most. Is a traitor requesting blueprints? Sure, he can have them, as long as he gets back to work. Is someone threatening murder if they don't get what they want? Again, a consideration has to be made about disruption to crew efficiency, and which path will lead to less of it. There is a revolution, and 2 out of 5 surviving heads are in robotics. Do you assist security, knowing this will lead to a prolonged battle and many deaths, or do you let the crew persuade you that calm will resume with less maintenance needed if you let them kill only 2 people off quickly? Several gangs arrive, and security is swiftly killed. Assuming the dominator doesn't actually cause more station maintenance (it's not really clear what it does), do you support the strongest gang to ensure a swifter, more bloodless coup and reduce strains on the station? A double agent convinces you that he has the capability to release the singularity at a moment's notice, and you are unable to make it more secure. Do you agree to help him find his target in exchange for avoiding greater damage to the crew and station as a whole?
My point is that you don't have to play this as a game of 'who can murder most'. The AI isn't an unwilling passenger in this lawset, forced to be buffeted around by the whims of others, but it doesn't have to be security either. It does what's best for station efficiency/maintenance, and then it does what's best for crew efficiency/maintenance. If that means stopping antagonists, sure, it can do that, but sometimes it may mean helping them as well. Yes, damage to the station might necessitate more maintenance, and yes, harm to the crew might require them to be replaced or cloned, but that is always better than it being destroyed outright or rendered totally useless. An AI with this lawset can choose to be pragmatic, assuming it is enforced well. It is neither a force for good or evil: it is, potentially, deliciously neutral.
