Difference between revisions of "AI War:AI Story"

From Arcen Wiki
Jump to navigation Jump to search
Line 15: Line 15:
  
 
Think of the AI as a distributed cloud network; it's the only way to efficiently have a database of that size.  Take Google as an example, for instance: they have hundreds of thousands of servers (or more), each with various data and processing capacity.  In order to run a search, the input of a large number of those servers are needed.  There is redundancy built in for any data of importance, but if all of the servers are down that have a particular datum, you won't get that datum in your search results.  The AI in AI War is fairly similar, in that all of the ships are not aware of everything in the galaxy at once. Note that this is in no way a technical limitation -- as you see in many RTS games, often the problem is the AI knowing too much, and thus seeming unrealistic.  In AI War, the realistic, gameplay-enhancing way was to simulate the AI as if it were a huge cloud network or bureaucracy; any other design and it wouldn't make much sense that it doesn't crush you immediately.
 
Think of the AI as a distributed cloud network; it's the only way to efficiently have a database of that size.  Take Google as an example, for instance: they have hundreds of thousands of servers (or more), each with various data and processing capacity.  In order to run a search, the input of a large number of those servers are needed.  There is redundancy built in for any data of importance, but if all of the servers are down that have a particular datum, you won't get that datum in your search results.  The AI in AI War is fairly similar, in that all of the ships are not aware of everything in the galaxy at once. Note that this is in no way a technical limitation -- as you see in many RTS games, often the problem is the AI knowing too much, and thus seeming unrealistic.  In AI War, the realistic, gameplay-enhancing way was to simulate the AI as if it were a huge cloud network or bureaucracy; any other design and it wouldn't make much sense that it doesn't crush you immediately.
 +
 +
[[AI_War:Fleet_Command]]
 +
 +
==What Does AI Progress Really Represent?==
 +
 +
'''Q:''' Why the generic term "AI Progress" instead of something like "AI Alert," "AI Alarm," or even "AI Disgruntlement" (kidding)?
 +
 +
'''A:''' AI Progress is the most general encompassing term, given its many interrelated meanings.  Some people have referred to it as the AI "rageometer," which is amusing but misleading.  The AI Progress is really a measure of the following combined factors:
 +
 +
- How annoyed the AI is at you (rageometer).
 +
- How big a threat it thinks you are (being frightened by things such as golems or warheads).
 +
- How ready it is to do anything about that (how much of its production potential is devoted to thwarting you).
 +
 +
From a design standpoint, the game could perhaps could have split out those sorts of things into multiple meters (AI Rage, AI Intimidatedness, AI Preparedness, etc), but in practice that likely would have been prohibitively complex without much added value to the game itself.
 +
 +
[[AI_War:Fleet_Command]]
 +
 +
==Why Does The AI Progress Go Up When The AI Destroys Key Structures Of Mine?==
 +
 +
'''Q:''' I am just wondering why the AIP goes up if a Golem is killed, or a human settlement is nuked? And similar structures I own that the AI would really benefit if I lost. I understand a Golem being repaired makes AIP go up, and the AI is in its right to get scared of that, but why when it dies?
 +
 +
'''A:''' First please make sure that you have read the above entry, about what AI Progress really means.  From that perspective, in these cases it's not anger, but rather the AI seeing you as more of a threat to itself.  To cite another animal example: Suppose you have a cat, and it has claws, but never attacks you.  Your estimation of that cat as a threat is very low.  But, let's suppose that cat then does something threatening, like arching its back or swatting at you -- even if it does not hurt you in any way, you still think of it as being more of a danger.  In many ways, this is why the Golems, even when dying, are considered a threat by the AI.  It thinks to itself: "Okay, ''that'' golem is dead, but these human build golems now, evidently.  So, logically, if I killed the one are they going to try something else that I don't know about yet?"
 +
 +
The other case, regarding the captive human settlements, again works best as a pet analogy.  Let's suppose you have a dog that you keep in the house, and he is known for trying to escape out the door when you open it.  That makes you take steps to try to contain him, and thus puts you more on guard with the dog not because of danger to yourself, but because you want to keep him contained and he is trying to thwart that.  With the captive human settlements and rebel colonies, those were presumably captive for a reason, and so if the AI then is losing control of them, that is cause for concern to it.  The dog is escaping, or got out and then got itself killed.  The AI has a lot of captive human settlements, and an uprising or an escape in one (especially if one is annihilated) might logically spawn rebellions or trouble in the others.  Hence, the AI's level of concern again rises -- just what are those pesky humans up to, and what might they do next?
  
 
[[AI_War:Fleet_Command]]
 
[[AI_War:Fleet_Command]]

Revision as of 12:39, 29 January 2010

How Does The AI Think?

Q: From a story sense, if the AI were smart, wouldn't it just kill me immediately with all those ships? I get why the current system makes for good gameplay, but if I were playing the AI side I would win in under 5 minutes.

A: This is a complex question, so let's break this down into parts:

Humans Have Already Lost And Are Now Considered Insignificant: The AI does behave realistically, if you assume that it is occupied doing something else largely. It is doing whatever non-war projects it fancies -- it's already won the war, for crying out loud -- and it doesn't really care about the remaining meatheads except insofar as it swats the ones that come up.

When ants invade your house, you kill them when you see them, you put out traps or poison to kill them as they come through, but you don't really spend much interest on them. They're ants! When small dogs and wolverines and bears start coming into your house, you react very differently. The AI Progress reflects the AI's estimation of your threat level moreso than anything else. If you get to the point where it sees you as a bear in its house, you had better be prepared to kill it very quickly or it will kill you.

The AI Is Busy Doing Something Else: In this specific scenario, the AI is also quite occupied with other pursuits. Again, the ant example is apt: you probably have a job, or go to school, or have something else occupying your daily attention even if you're retired. You have a family, friends, and other social interests and obligations. You're trying to make a mark on the world, and provide for yourself and yours, same as the rest of us. The AI is similarly engaged outside the galaxy, although what exactly the AI is up to out there won't be revealed until a future expansion. One side effect of the AI's activities outside the galaxy is a lot of new weapons and technology, though: all of the core/ mark V ships result from their extracurricular pursuits, as do most of the AI superweapons such as attrition emitters, exo-galaxy wormholes, ion cannons, and so on.

The AI Is Not A Centralized Intelligence: The AI is not omniscient about everything that is going on with all of its ships. Generally it knows that humans are nearby when humans border one of its planets -- hence the AI going on alert at those planets -- but other planets out in the galaxy may have no idea there is a fight going on at all. The AI Progress measures the AI's global awareness of you, as well as its global estimation of you, and that creeps upwards over time as you damage it to an increasing degree (or "scare it" with massive weapons of your own, such as warheads or golems).

Think of the AI as a distributed cloud network; it's the only way to efficiently have a database of that size. Take Google as an example, for instance: they have hundreds of thousands of servers (or more), each with various data and processing capacity. In order to run a search, the input of a large number of those servers are needed. There is redundancy built in for any data of importance, but if all of the servers are down that have a particular datum, you won't get that datum in your search results. The AI in AI War is fairly similar, in that all of the ships are not aware of everything in the galaxy at once. Note that this is in no way a technical limitation -- as you see in many RTS games, often the problem is the AI knowing too much, and thus seeming unrealistic. In AI War, the realistic, gameplay-enhancing way was to simulate the AI as if it were a huge cloud network or bureaucracy; any other design and it wouldn't make much sense that it doesn't crush you immediately.

AI_War:Fleet_Command

What Does AI Progress Really Represent?

Q: Why the generic term "AI Progress" instead of something like "AI Alert," "AI Alarm," or even "AI Disgruntlement" (kidding)?

A: AI Progress is the most general encompassing term, given its many interrelated meanings. Some people have referred to it as the AI "rageometer," which is amusing but misleading. The AI Progress is really a measure of the following combined factors:

- How annoyed the AI is at you (rageometer). - How big a threat it thinks you are (being frightened by things such as golems or warheads). - How ready it is to do anything about that (how much of its production potential is devoted to thwarting you).

From a design standpoint, the game could perhaps could have split out those sorts of things into multiple meters (AI Rage, AI Intimidatedness, AI Preparedness, etc), but in practice that likely would have been prohibitively complex without much added value to the game itself.

AI_War:Fleet_Command

Why Does The AI Progress Go Up When The AI Destroys Key Structures Of Mine?

Q: I am just wondering why the AIP goes up if a Golem is killed, or a human settlement is nuked? And similar structures I own that the AI would really benefit if I lost. I understand a Golem being repaired makes AIP go up, and the AI is in its right to get scared of that, but why when it dies?

A: First please make sure that you have read the above entry, about what AI Progress really means. From that perspective, in these cases it's not anger, but rather the AI seeing you as more of a threat to itself. To cite another animal example: Suppose you have a cat, and it has claws, but never attacks you. Your estimation of that cat as a threat is very low. But, let's suppose that cat then does something threatening, like arching its back or swatting at you -- even if it does not hurt you in any way, you still think of it as being more of a danger. In many ways, this is why the Golems, even when dying, are considered a threat by the AI. It thinks to itself: "Okay, that golem is dead, but these human build golems now, evidently. So, logically, if I killed the one are they going to try something else that I don't know about yet?"

The other case, regarding the captive human settlements, again works best as a pet analogy. Let's suppose you have a dog that you keep in the house, and he is known for trying to escape out the door when you open it. That makes you take steps to try to contain him, and thus puts you more on guard with the dog not because of danger to yourself, but because you want to keep him contained and he is trying to thwart that. With the captive human settlements and rebel colonies, those were presumably captive for a reason, and so if the AI then is losing control of them, that is cause for concern to it. The dog is escaping, or got out and then got itself killed. The AI has a lot of captive human settlements, and an uprising or an escape in one (especially if one is annihilated) might logically spawn rebellions or trouble in the others. Hence, the AI's level of concern again rises -- just what are those pesky humans up to, and what might they do next?

AI_War:Fleet_Command