Well, the biggest limitation of the way the DW handles AI, is that it isn't possible for a platform to reference two other platforms relative to each other.
In other words, if there is a distant target and a friendly or neutral in between, there is no way to say to the objective platform "don't fire on that target because there is a potential blue-on-blue". In DW, the AI going to fire, and the faith is that the automatic presets put on the weapon will prevent the friendly fire... obviously, this doesn't always happen.
I've been thinking about how to get around this, but no luck so far...
Another limitation is that the doctrine system appears to merely provide the AI platforms with "suggestions" on what they should do in terms of weapons use and how they handle contacts. Very few of the doctrine commands actually command the AI to do something directly... it's more like it puts the idea of doing it on their proverbial radar, and then the AI core in the NavalSimEngine determines if they are actually going to do it.
I'd imagine the contract versions of DW have a much more reliable way of directing AI behavior...
Cheers,
David