If $\hat{\theta}$ is an unbiased estimator of $\theta$ and $T$ is a sufficient statistic for $\theta$, what does the Rao-Blackwell Theorem state about the estimator $\tilde{\theta} = E[\hat{\theta}|T]$?