I believe that true art should remain scarce. A single author cannot produce thousands of genuine artworks in a year; if art is to carry meaning, it must come from time, attention, and commitment. Quantity is oftentimes an enemy of quality and uniqueness.
For AI art, this means that a generated work should not be rushed for publication as the immediate product of one AI prompt, however well crafted, let alone as a batch of variations of the same. To be recognised as art, it should involve substantial manual labor and a series of artistic decisions — editing, curating, refining — that reflect the author’s creative involvement.
I also think that an artist should maintain a lasting relationship with their works, including those that make use of generative AI. An artwork is not something disposable, but a piece of a broader vision that develops over time. The author should be able to remember each of their creations, no matter how long ago they were produced.
Finally, every artwork should show something unique: a personal style, a technique, or a perspective that distinguishes it from others. Without this distinctiveness, there is no real authorship.
These are the principles I have come to recognise after two years of my involvement in generative AI imaging. I would be interested to hear how others see it: what, in your view, makes a digital or AI-assisted piece of work truly art?
(A disclosure: I may not be always following this manifesto to every point in my own practice, but I am striving to. Practically each image you see here in my portfolio is a result of thorough selection among tens of generated variants, and more often than not I edit a visibly imperfect work before publishing, by way of inpainting. And I am gradually converging on a few distinct styles and techniques of my own. Or so it seems.)