This is possible with perfect information. However, its a lot more tricky in reality. You don't know which defending ships have launched the missiles. You don't know if they have full or nearly empty magazines. You don't know if they are damaged. You don't know what other defenders might be in the same location, or may have been last time you fired but not this time. You don't know if there are beam defenders in the same location that simply haven't fired yet because you didn't penetrate the missile defences, etc..
You could easily end up with NPRs refusing to fire because they don't think they can overwhelm the defence when in reality the defenders are damaged or out of missiles or the ship that was really doing the defending is no longer with them.
You don't need to have perfect information of the defenses. If you don't have information you can use trial and error in order to increase the amount of information you have available.
What a player will commonly do is launch a single salvo (or a few salvos if enemy AMM have long range) and watch the effect.
Some interesting information I use to decide (that the AI might also find useful) are:
- Total known enemy fleet tonnage, compared to own tonnage of AMM armed ships (including their ammo status).
- Interception chance of enemy AMMs.
- Interception location (time left to impact when first missiles are intercepted)
As you can see the last two can be determined from launching a single "probing" salvo.
If you know the chance of interception and how many ticks your missiles have left to travel after first interception you can get a quite decent estimate on if you will be able to penetrate the AMM umbrella.