Isn’t it P(A|C) = P(B|C) = 0 where C is both slits are open? P(A)+P(B) is still > 0. The condition C modifies the marginal probabilities of A and B, does it not?
This was a very effective way to describe it, I've read a great deal about QM and you made it click much better.
Amplitudes are like probabilities that can destructively interfere when added. With traditional probabilities, if P(A) > 0 and P(B) > 0 the sum of the probabilities P(A)+P(B) cannot be zero. However, we needed to model observations like the two slit experiment where we observed that a location on the target that had positive probability of being hit by a photon in the two single-slit-only situations, that became zero when both slits were open.