The Artist’s Hand: AI Art Pioneers That Existed Before the Boom

The artist’s hand refers to the evidence of an artist’s personal and unique touch left in a work. This can be seen in the specific brushstrokes of a painting, the modeling of a sculpture, or even the overall emotional quality of a piece. The proof left behind reveals or provides insight into the artist’s role in creating the art. But can the artist’s hand emerge in AI art? Aaron Hertzmann first mentioned the artist’s hand to me in a chat thread (see below). Hertzmann is a principal scientist at Adobe Research and he specializes in computer graphics, computer vision, and machine learning. Hertzmann argues that computers do not make art; people do. He consistently rejects claims of machine creativity, emphasizing that art is a social phenomenon and that AI algorithms, despite their impressive capabilities, are tools used by human artists.

In 2022 Hertzmann posted a video of Harold Cohen discussing his drawing algorithms in the late 1980s. Hertzmann frequently references Cohen and his AI-driven software in his papers, articles, and talks to illustrate historical perspectives on AI in art. He uses Cohen’s experience to highlight the ongoing debate about authorship and the evolving criteria for what constitutes art as technology advances. The connection between Hertzmann and Cohen is rooted in their shared domain of AI and art, with Cohen as a pioneering practitioner who created AARON, and Hertzmann as a contemporary AI scientist and theorist who analyzes Cohen’s legacy and the broader philosophical questions raised by generative algorithms.

Luigi Russolo’s Intonarumori (1913)

There are a few other artists of note that pre-date Harold Cohen. In a previous post, I wrote about Italian Futurist painter and composer Luigi Russolo who advocated for the use of “noise-sounds” to break free from the limitations of traditional musical instruments. Russolo’s concept extends beyond sound to include visual “noise,” such as the use of visual noise to generate AI images. Russolo built different instruments to perform music he outlined in his The Art of Noises manifesto written in 1913. The instruments or boxes had various types of internal construction to create different types of noise music.

Marcel Duchamp’s Endgame (1930)

“Endgame” refers to artist Marcel Duchamp’s profound, lifelong passion for playing chess, viewing it as a purer, cerebral art form than painting, and a central metaphor for life, art, and the avant-garde’s shift to conceptual ideas. Duchamp eventually retired from art to take up chess. He achieved the status of grand master and in 1931 wrote a book on endgame strategy obliquely called Opposition and Sister Squares are Reconciled.

In the 1990s, when World Chess Champion Garry Kasparov played against IBM’s Deep Blue supercomputer, many in the chess community feared a victory for the machine would deliver a fatal blow to their 1,500 year old game. Note: Kasparov was eventually defeated by the computer in 1997.

At the time, computer technology was spreading globally and many chess veterans reasoned that if the best human player was shown to be inferior to a machine then interest in playing would wane and sponsorship would disappear. They urged Kasparov not to take part in the spectacle. — Martin Rand

However, machines learning chess did not discourage humans from playing. In fact, professional chess has grown exponentially in popularity over the years. Machines didn’t replace humans in chess. It changed how humans played the game. The same can be said for chess and later machine learning/AI. For Duchamp, art, like chess, was about following rules. However, art usually isn’t about following the rules.

Sol LeWitt: The First Prompt Engineer

In the 1960s, artist Sol LeWitt came up with a new way of creating art. Lewitt was known for creating “procedures” or “systems” that anyone could use to create art. He called these procedures prompts. LeWitt provided artists with “structures,” or detailed assignments they could follow in order to produce pieces of art which embodied and effectively communicated their concept. He sought to make art accessible and encouraging people to create and view art without being confined by traditional formats and expectations.

Sol Lewitt’s drawings were made right on the wall. The people drawing them use a set of instructions written by the artist. The instructions are open to interpretation: people following them have to decide where the lines or shapes should go. Though the instructions are clear in a sense they are not so precise as to result in an exactly replicable outcome.

On a wall surface, any continuous stretch of wall, using a hard pencil, place fifty points at random. The points should be evenly distributed over the area of the wall. All of the points should be connected by straight lines.

Other artists such as Amy Goodchild have repurposed LeWitt’s prompts using AI-generators such as ChatGPT and, for this entry, I used one of his prompts in Midjourney. My prompt:

A portrait of nina simone based on fifty random points that are evenly distributed. All of the points should be connected by straight lines.

[Harold] Cohen didn’t give AARON instructions like, “Draw a tree.” He gave it concepts — how things are shaped, how they relate to each other, and how to represent a world that AARON never truly saw, but understood through abstraction. AARON, in a sense, was a world-builder before world-building became a buzzword. — AIU

When artists today experiment with generative AI tools, they should be aware of the pioneering spirit of those who came before the ‘boom’. They are the people doing what Maya Ackerman refers to as “shadow work”:

It is the ongoing practice of identifying, acknowledging, and reintegrating the parts of ourselves we’ve disowned… Shadow work isn’t about perfection or performance. It’s about seeing clearly.

It’s about participation. Pioneers are not concerned with copying artist’s works. They have a major role in instructing machines on ways to create artworks that show what is possible with algorithms and machine/deep learning, sometimes going beyond what many traditional artists have the tools to do. According to AI Innovations Unleashed, in today’s landscape of copyright lawsuits and data-scraping controversies, AARON (and other pioneering artists’ works) provide us with blueprints for ethical machine creativity.