I think @Kanitatlan
has the best explanation of the problem. Specifically, when the AI is on the attack and has a success, it then moves its forces to help the other attacks that are not succeeding, instead of pushing through where the success happened. If the AI did stop the failing attacks and reinforce the one that succeeded, then the AI's loss ratio would drop dramatically as it encircles or unwinds the player's front line.
As far as comparing the 30 to 1 losses in game, this may be evidence of the problem Kanitatlan has described. In real life, a human general would call off failed attacks that are trading so badly. Sure, for short periods of time 30 to 1 trades can happen, but they are stopped before they become an operational problem. In the game the AI does not stop these bad trades. As Kanitatlan said, instead the AI stops the attack that succeeded and reinforces all those attacks that are trading badly. If a human general did this over and over for years, he would most likely suffer the same bad trades.
So, do we fix the AI creating its own problem of 30 to 1 trades, or do we create a combat system that rewards such horrible tactics?
As for the OP's idea of keeping org damage the same, but possibly increasing damage to strength, that is an idea that could use more discussion. My first thought of strength is that reinforcement of strength happens so fast, that it can feel like we are not fighting the equipment of the divisions in the battle, but the inventory of all equipment in storage. For the AI especially, it comes down to a world wide equipment grind it cannot win, unless it is outproducing the player by a larger ratio than it is losing equipment.
Maybe the strength reinforcement can be slowed down so the attacker has a chance to take advantage of the local losses of equipment. It would make three to one fights more 'real' as more and more defenders and attackers see their orange strength bars go down and the side that started with more divisions has an advantage as that side will have full strength divisions still to attack with as the other side runs out of full strength divisions.
But it would take more than that. The loss of strength also may need to reflect in the stats of the divisions. From what I understand now, loss of strength does not immediately reduce the stats of divisions in combat. If this is correct, and I may be wrong, then this would need to be fixed or we are right back to just grinding each other's storage of equipment down, instead of the divisions in the battle.
Addressing strength and how fast it is reinforced could make reserves important, which is another thing the AI will need to learn.
I hope someone can gives us better insights on the issue than I can give. This is an interesting discussion.