In much the same way people think of digital storage as external memory, I think of generative A.I. as external imagination. Of course, human memory doesn’t work like a hard drive, and LLMs don’t work like our imaginations. But as a guiding metaphor, it seems to work well for identifying good/bad use cases.