| Forums-->Queries and help-->AI again improved?|
1|2Back to topics list
|Author||AI again improved?|
|There us quite a bit difference in it. Can't understand why you disagree|
It's reasonable to assume that AI has some threshold upon which it bases its decisions.
You said it yourself that you have seen a smaller stack move away to make room for a bigger stack. I have seen it myself, where a stack attacks till certain point, when it becomes small enough, at which point it moves away. I didn't pay close attention when it makes this decision, but I'm 100% sure that it doesn't move away as soon as it becomes smaller than another stack. So clearly there's a threshold where it decides that it's better for the bigger stack to go at it. Is it when it's half, a third, a fourth, a fifth of the other stack?
It's quite conceivable to me that the threshold be any of those even for a smarter AI. All I was saying that maybe the stack size wasn't different enough for the AI to make the decision to switch. I didn't dispute Meshy's claim. I was just saying that that battle was not convincing enough (and of course RecorderPT's was) . There's definitely room to guess. And yet you act like you know exactly how AI thinks.
I guess you must be really smart.
|To be fair, in my opinion both sides have a point, though the passive aggressiveness wears a bit thin.|
One thing to be clear about is the fact that the battle by teh person who opened this thread the decision by AI was not to move away from a position of direct contact and attacking, but prior to engagement. This may be a factor.
Equally it may be an artifact of a different AI interaction which is likely familiar to people whereby a stack that would die in retal would not attack when the total army strength is still high and possibly sufficient in its judgement to win.
As such in the first battle it was a stack of 8 horses against some patriarchs with an attack build. In the other two they were facing gargoyles of a magic build. As such the relative strengths of the stacks in melee combat are differently matched, and so it is perhaps reasonable that the AI would not judge the need to move away.
At the end of the day all we have is educated guesswork, no one persons guess can trump anothers per se, though some may be based on more robust evidence and less assumptions, though without knowing the background coding, such an elusive thing as proof is a bit more distant.