Let me tell you why war is so common in a typical game.
It's because human players aren't bound to the patterns of AI behavior. AI leaders, on the other hand, are bound to the patterns of their behavioral profile defined by the AI programming.
If you knew that Brennus would never attack you if he's Friendly, that's a one-sided advantage that the human player has vs. the AI. Add several of those AI leaders into the mix, and you potentially have several AI leaders that won't attack the human because they are Friendly to him.
In practice, this puts the AI leaders at a disadvantage compared to the human, because the human isn't bound by the patterns followed by each AI leader, but they play such that the human is bound by them.
In other words, if the human doesn't start a war, then he loses the opportunity to exploit this aspect of AI weakness--which is the weakness of being bound to AI leader behavioral profiles. If the human doesn't war, then he becomes more like the AI leaders by more closely conforming to the AI's expectations of the human player's actions. The result of this is an effectively stronger performance from the AI in decision making.
Therefore, to make war a high priority in gameplay, as far as the human player is concerned, would be to minimize the capability of the AI to correctly anticipate the human player's actions. Collectively, this minimization of AI strength with respect to all rival leaders gives the human a distinct advantage relative to the AI leaders, when he adopts the war-heavy strategy.