It is a problem because people will talk about what AI can do implying that an LLM can do that thing, making it seem like a pure LLM can do almost anything. On the other hand people will say AI will never be able to do X because an LLM can’t do that thing well natively. AI has become too vague of a term to be useful.