One thing I really love above the new AI is it's willingness to declare war if you're already at war with someone else, especially someone very far away. For example, playing as the Antigonids who already conquered all of Persia and Egypt, when attacking Kush, Maurya, Rome, and Armenia all eventually declared war on me. This is a great strategy and forces the player to keep reserves instead of committing an entire empire's forces in one front, which is realistic and good for balance.
However, the AI really sucks, and it is super easy to defend against AI even with a much smaller force. Their primary weakness is a failure to concentrate on the war goal. For example, when Maurya declared war, their troops were all spread out and sieging aimlessly, while, only a few cohorts were sent to the war goal. This makes it easy for the player to defend the war goal, gain ticking war score, and eventually secure white peace (or better).
As an AI programmer, I know that it's hard to create smarter AI. But I think adding some kind of "importance" metric to provinces to guide AI army movement could significantly improve the AI (this logic is similar to the province occupation goals that you can issue to allies in EU4). The importance metric doesn't have to be very sophisticated: highest importance to the war goal, and decaying importance based on distance from the war goal would encourage targeting the war goal and consolidating gains there before branching out. I think even this logic will make defending against AI attacks much more challenging. Of course, this can be made better by adding more importance to provinces with claims, or to more populous provinces (more slaves, more damaging to the opponent's economy). If this logic already exists, I think it needs to be tuned to target the war goal with higher priority.
However, the AI really sucks, and it is super easy to defend against AI even with a much smaller force. Their primary weakness is a failure to concentrate on the war goal. For example, when Maurya declared war, their troops were all spread out and sieging aimlessly, while, only a few cohorts were sent to the war goal. This makes it easy for the player to defend the war goal, gain ticking war score, and eventually secure white peace (or better).
As an AI programmer, I know that it's hard to create smarter AI. But I think adding some kind of "importance" metric to provinces to guide AI army movement could significantly improve the AI (this logic is similar to the province occupation goals that you can issue to allies in EU4). The importance metric doesn't have to be very sophisticated: highest importance to the war goal, and decaying importance based on distance from the war goal would encourage targeting the war goal and consolidating gains there before branching out. I think even this logic will make defending against AI attacks much more challenging. Of course, this can be made better by adding more importance to provinces with claims, or to more populous provinces (more slaves, more damaging to the opponent's economy). If this logic already exists, I think it needs to be tuned to target the war goal with higher priority.
- 3
