When answering questions about light, I see that we conveniently shift between wave and particle nature of light to match the answer-- isn't this really cheating?
Or, is it the principle that the observation changes the nature of the object?
You wouldn't think it's cheating if you understood the "wave" involved to be NOT a physical wave in some ether-like medium, but a measure of the information we have about the system (technically, a wavefunction -- the complete description of the system).
When there is no detector to access that information, the wave is spread out in space, more wave-like because of its spatial extent and displays all of the typical wavy behaviour of interference and diffraction. If you now have a detector that measures position, it will localize the wave in space to such a small region that it will appear to behave like a particle, but really still is a wavepacket -- albeit very sharply peaked around the measured position.
The details of how detectors and measurement devices are able to do this is partially described by the theory of quantum decoherence (some would argue, fully described). But I hope it's clear to you that the wave-particle duality is not some kind of ad hoc prescription; it's a natural consequence of asserting that the wavefunction is a complete description of the system.
Yes, it's cheating. Treating light as either waves or particles is cheating. If you don't want to cheat, you should do everything from first principles in quantum optics and never mention either waves or particles (both of which are only approximations to the true nature of light). Unfortunately, to do that you have to pretty much throw all your intuition out the window. Cheating is much easier.
The truth is that light (like other things) isn't a wave or a particle. It's another kind of a thing, namely an excitation of a quantum field. Under some circumstances, excitations of quantum fields behave like waves; under other circumstances, they behave like particles.
It's not right to say that we switch descriptions "to match the answer." Rather, we switch descriptions to match the question! Certain experimental setups and measurements are more naturally analyzed by thinking about things as waves; others are better analyzed in the particle framework. There are good reasons for using one or the other in any given situation. In effect, both wave and particle pictures are approximations that arise as limiting cases of the correct, exact theory. If you study how those approximations arise from the exact theory, then you can understand when one or the other is appropriate.
One final note. I hate language like "observation changes the nature of the object." The nature of light is always the same: excitation of a quantum field. Coincidentally, I ranted about this elsewhere a bit recently.
To add a lighter note, a reminder of the four blind men and the elephant story:
What is an elephant like, asks the first.
It is like a snake says the second, holding the trunk.
You are wrong, says the third, It is like the trunk of a tree
Ha ha ha says the fourth, what a great imagination you all have, to see this great big wall as snakes and trees leaning on the elephants side.
Strange! you are all confused, said the first, I have caught a long piece of string (wink, wink)
At least we have a dual ambiguity only:) .