A.I., art and human beings

A-digital artist musings on AI-image generation

Sep 30, 2022

There has been lot of discussion in the last weeks about images generated by AIs like Dall-E or Midjourney. It re-ignited for the nth-time questions like “What is art?”. While defining art might be complex, at least one should consider what may be (or may not be) art. For example: is a Midjourney image art? If we only focus on the “external” aspects of art, such image might well deceive art experts (well, maybe not today, but when AI is involved, dismal improvements are usually just a matter of time…)

If an “art expert” (or simply ourselves) can’t distinguish one “good” Midjourney image from one “good” image created by a human (in terms of artistic qualities), maybe the AI image should be considered “artistic” (from the “contemplation point of view”)

To me, there are other points of view that should be considered too: process and production.

Art process

I believe that the process of creating an artwork is important. If we only need to push a button of a machine, is the result Art? A photo by itself is not art. Artistic photos are the result of the viewpoint of the photographer: composition, framing, selecting what is in the photo and what not… Even though using A.I. tools require skills and creativity in terms of prompts, they just produce pseudorandom images. The prompter decides which one is interesting. To me, only “choosing” is not enough for Art. There should be some intention (even if vague), exploration, experimentation…

This leads to another thread: A.I images may not be by themselves art, but such AIs may become another tool to help artists in creating artistic images - in a similar way as other digital tools, like illustration or photo-editing apps.

I de Inteligencia Artificial (“A for Artificial Inteligence”)
Rosalía - “Abcdefg”

Make good art

It has been a while since the beginning of the “Do it yourself” movements. Nowadays there are so many tools in so many fields that “amateur” production of culture has never been easier. This, combined with social networks, has led to the current situation where billions of bytes of media are created (and consumed) constantly (as you may have noticed, in this post I am focusing on Visual cultural products). To me, consuming culture has a “passive” quality - even though art forms like Comic require a great deal of “implication” by the reader. Producing culture - even a “bad”, simple drawing- allows oneself to be involved in the expression, leave a mark. As mediator in illustration / comic workshops, I try to motivate everyone I meet to allow herself to create culture: a short poem, a small picture, a brief play. I believe that creating art (specially when our body is involved and allowed to interact with the physical world) is powerful and helps greatly in growing as human beings. Neil Gaiman contributed a beautiful “Make Good Art” speech that you can read.

Challenges for AIs

managing ellipsis - how can AIs learn to manage what it is missing in the training datasets? This is specially important in comics, where ellipsis happens continuously in the gap between panels… and, in general, I believe that Art mostly deals with what we do not know (as Lynda Barry wonderfully reveals in “One! Hundred! Daemons!”)
Dave McKean points out that he considers drawings one of the main challenges for visual AIs. I agree that a good drawing is not just “making a photo” of something with a pencil. Using your personal point of view you create a “stylisation”: remove some details, maybe add others, highlight some parts… and everything with simple elements, like dots and lines.

What do you think the implications of AI tools will be? What has been your experience with Midjourney? What does art offers you (whether consuming it, or producing it)?

Atelier Tranquilinho

Discussion about this post