Actually I am confident exactly because I do know how LMMs work, and your comment fails to address the issue at all. Such models can't tell you anything useful about the probability that a particular statement is accurate.