Please check this Rabi oscillations image1:
Now, let's say I want to create a superposition of E1 and E2, so I started an EM field tuned to a frequency related to E2-E1. But, let's say there is another energy difference E3-E1, as per the above image there is still some probability (though it will never be 1) that E1 to E3 transition might occur. Isn't it? Or is there something wrong in my understanding?