Essays examining subtle epistemic failures in contemporary AI systems, especially cases where models sound reasonable while quietly substituting argument and avoiding responsibility to truth.