I've generated a lot of images with these tools, but they're not art.
The problem with image generation is that, unlike a camera which replaces the choices that a painter would make while painting a scene with the choices a photographer makes while shooting it, image generation robs an artist of the ability to make choices while constructing their work and doesn't replace them with anything substantial.
Now you might be tempted to say that engineering the prompt lets you make choices while generating images. It does so only in the most surface-level way possible, alienating the "artist" from 99% of what goes into creating an image. All of the choices that a painter would make while the paint is hitting the canvas are instead made by the algorithm, and they are made specifically to copy the choices artists in the past have made, not to come up with anything novel or unique. Then the person who wrote the prompt views the output and decides if it's "close enough" to what they had in mind or not, without exercising artistic control over the process at any point save for the very beginning and end of it.
That is not to say that you can't take a generated image and start making art with it, but every pixel that was generated that you don't change is a choice that the bot made that you didn't, a tiny bit of alienation from your own work that you have invited into the artistic process, and frankly that cheapens it.