Humility has to go both ways then, we can't claim that LLM models are actually (or not actually) AI without qualifying that term first.