One can hope. I've seen some real impressive use of language to generate very different looking images. Friends of mine who are designers see this as how their jobs will work in the future - they'll use different inputs to guide the AI in how the art should look, and then do finish work on the generated images.