The same feature that makes LLMs unreliable and dangerous for factual work also makes them useful. The real question becomes how to use AI to take advantage of its strengths while avoiding its weaknesses. To do that, let us consider how AI “thinks” creatively.
