AI images – Modern Art Exhibit

A modern art exhibit envisioned by me and Midjourney, as a (kind of) homage to Yayoi Kusama.

I keep noticing a few things about AI-generated images.

THE COOL THINGS EXISTENCE CAN DO

For me, AI-generated images are not so much about me, and much more about humanity as a whole. It’s very clear that these images are based on the collective artistic production of humanity. And that’s a reminder that all art is that way, even if it’s expressed through a local and temporary individual.

When I show the AI images I create with Midjourney, it’s not so much “see the cool things I can do” but “see the cool things we can do”. It’s about the cool things we as a society, humanity, and existence as a whole can do.

Really, it’s about the cool things existence can do through and as its local and temporary expressions we call humanity, culture, society, individuals, and technology.

BLURRED BORDERS BETWEEN AI AND HUMANS

Some assume there is a clear boundary between AI-generated images and human-created images.

Yes, it’s good to make that distinction.

And no, the boundary is not as strong as some like to present it. The AI is created, designed, trained, and modified by humans. It’s also trained on a specific set of human-created images and is dependent on new human-created art to continue to evolve and get new impulses. A human needs to envision the image, design and refine the prompt (which requires some skill and knowledge of visual arts), and select and edit the image. And many use AI images to inspire hand-made images and art.

The AI is created, trained, and modified by humans. It depends on human art to get trained and find metaphorical inspiration. It influences human-created art. The distinction is not as clear as it may first appear, and there is certainly no fixed or strong boundary anywhere between AI-generated images and hand-made images and art.

THE LIMITATIONS OF AI

Any AI that generates images – and other things – will obviously have limitations.

Midjourney (MJ4) can do some things very well, and other things not so much. It is good with faces but has trouble with hands and the number of fingers. It knows some styles very well and is less familiar with other styles. (This goes for both general styles and the style of specific artists.) It’s familiar with some geographical locations and less familiar with others. It is good with individual objects and has trouble producing images with several unique objects with specified relationships. And so on.

These start-up problems will be fixed, but any AI will always have limitations.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.