> However, in that framework, in many cases you're still often left with uniform priors, which then reduce to frequentist statistics.

Several commenters have made this claim, which seems to be true only for maximum a posteriori (MAP) estimates. Frequentist methods do not construct a distribution over parameters.