LLMs can't solve such tasks from zero either. Self attention in transformer networks is decidedly not Bayesian, but self attention doesn't model truth or truth

LLMs can't solve such tasks from zero either. Self attention in transformer networks is decidedly not Bayesian, but self attention doesn't model truth or truthiness either. Transformer self attention models only model frequency of observation in the sample, not truthiness.