I've just used Midjourney. It generates beautiful pictures.
They are beautiful, but not art. Neither of the ones who asked the AI to do that nor the software made any artistic expression. They are, at their best, mere spectacles.
Art requires a certain awareness of things. You can consider art as an imitation of reality or a thing of reality itself, but you must know what the things are. It is not enough to see a thousand men only visually to make a portrait of a man uniquely artistic. One has to look into their soul (or in the lack of it).
That is where any AI based on machine learning of our time fails. With my prompt to imagine “death most banal”, the bot generated the first image. It neither understands what death is (no one does fully, but it does not at all), nor what is banal. A death of ultimate banality is a concept we cannot expect it to understand.
For the second image, I asked for “Bertrand Russell's dream”. It produced three portraits and the 2nd image here. Without the knowledge of who Russell is, one cannot expect anything more than this.
Art is in the artist's life, as much as it is in what the artist produces. I'm waiting for an artist machine yet to come. But, I doubt, with its artistic sense of freedom, will it ever take orders from us.
Susan Sontag disagrees with the idea that art is an imitation (which probably was originally). Instead, she insists that arts should be considered on its own merit. For more, see this highlight from the essay On Style. ↩︎
John Searle has works in this field. He proposed the term strong AI for AIs capable of this kind of awareness. Unfortunately, that is not the case for any AI currently available. See more on this Wikipedia page. ↩︎
The Starry Night সেকাল-একাল ↩︎