This paper by Carlip http://arxiv.org/abs/gr-qc/0108040 is a good, relatively nontechnical explanation of why it's hard to reconcile quantum mechanics (QM) with general relativity (GR).
GR says that spacetime is a real manifold with a semi-Riemannian metric. QM says that the possible states of a system form a complex vector space.
If you naively try to combine these two ideas, it's hard to make sense of the result. Given one manifold-with-metric $M_1$ and another one $M_2$, what would it even mean to talk about the linear combination $c_1M_1+c_2M_2$, where $c_1$ and $c_2$ are complex numbers? The spacetimes $M_1$ and $M_2$ do not have any built-in way of matching up points in one with points in the other. The two spacetimes don't even need to have the same topology. In quantum mechanics, we would also have the Born rule, which says that $|c_1|^2$ and $|c_2|^2$ have interpretations as the probabilities of outcomes of measurements. It's not clear what these probabilities would mean in this context.
So should spacetime be described at the Planck scale as a real manifold, or if not, then what? Straightforward application of the fundamental principles of the two theories seems to lead to nonsense answers. We really don't know.