Asymptotic giant branch predictions: Theoretical uncertainties