Joseph Munisteri, aka Butterflies in Space Joe, is an artist, writer, and technologist based in New York. Each piece in Munisteri’s latest collection “Butterfly Space Opera” was created using various forms of Artificial Intelligence, inspired by classical modern artists and nature.
Over the next few weeks, we will be discussing the rise of AI Art in the NFT Space. This week, our Community Manager shares an op-ed about the legal and ethical reasons people are opposed to the new art form. Check back next week for Part 2 where we will discuss why it could be considered a new part of the art canon.
AI ART – CREATION OR REPLICATION?
PART 1 – REPLICATION
While fully computer generated art can be considered relatively new to the NFT Space, artists have been using functional algorithms to create art based on strict rules and artistic disciplines for the last half a century. Like any new medium, there has been a lot of pushback and debate surrounding AI art in the community. The argument is over whether or not these algorithms create art or simply replicate art based on their programming. This week, we dive into a brief history of AI and explore the opposing camp.
One of the earliest documented forms of utilizing computers to generate artistic images dates back to 1973, with the introduction of the AARON [non-acronym] project by prolific painter and engineer, Harold Cohen. This project is considered one of the longest standing and maintained AI algorithms in history. Around the time of the program’s inception, Cohen’s work was spreading internationally; his art focused on the abstract, biomorphic forms, and he frequently worked with colors that closely mirrored the pop art movement that was sweeping the creative world.
Cohen’s goal was to use AARON to produce drawings that followed a specific set of rules he had created, similar to the traditional artist studio assistants model. He continued to develop and refine AARON for the rest of his career, but the program maintained its core design of performing tasks as directed by him. While the early output of the system was rough in a primitive sense, over time Cohen would take the time to color select pieces by hand. Together, Cohen and AARON produced thousands of images at varying scales- from 8 ½ x 11 sheets, to massive murals. Notable to this method was Cohen’s constant input of rules, and involvement in rounding and personalizing the final works. This method keeps the artist in control, and allows for a sense of human creativity to blend with a mechanically reproduced image. This differs slightly from the more modern forms of AI art that are widespread and recognizable today.
Most of the AI artworks that have emerged over the past few years have used a class of algorithms called generative adversarial networks, otherwise known as GANs. They’re considered to be “adversarial” because they are two sided functions. One generates a random image, and the other has been taught, in a manner of speaking, to judge the images based on specific inputs to assign it to specific categories. Another program, AICAN, takes the use of these adversarial networks a step further with the end goal of assessing how close a generated image is to its pre-curated reference pieces. The program has also been written to avoid creating images too similar to the assigned source. AICAN attempts to create unique art based on these guidelines, but it is limited by the constraints of the algorithm.
The works these programs generate are built upon predicated artworks that were influenced by the human condition. Without the existing art, these heterogenetic networks would be unable to create in the sense that we as humans understand creativity- computers lack the psychic structure necessary for such feats. There can be no assigned or understood meaning on behalf of the creator. In essence, the feelings and emotions that have inspired creatives are non-existent. Where do we draw the line without the element of human consciousness?
Instances of the ongoing battle to define the limitations of using other’s art as references and defining the protections associated with doing so are plentiful. As recently as 2019, the US Copyright Office handed down a ruling that an AI-created image lacked the human-authorship that could grant it the ability to be protected by copyright. As the law currently stands, only works created of intellectual labor can be provided such protections- and the USCO has continually maintained that non-human expression will remain ineligible based on case precedent. Similar suits have been brought to the US Patent and Trademark Office and the UK Intellectual Property Office with the rulings aligning with USCO standards. From a legal standing point, AI Art is recognized as the product of algorithmic input rather than the creative power of the human mind.
Traditional art, independent of a specific medium, is typically valued based on knowledge of whether or not it was produced by a favorable artist, or using a specific technique that results in a desirable outcome to a collector. Economists and psychologists who work to define the value in art agree that this can be summed as the terms of a piece’s authenticity. Authenticity, much like the art itself, can also be perseverated subjectively as the rise of “motivational” authenticity (otherwise known as artistic integrity) often comes into play when assigning value to work. This concept focuses on the “why,” and this is where skeptics of AI art find a foothold in defining their argument.
Altering existing art and claiming the resulting expression is owned as intellectual property by the individual who used the original piece as reference calls the ideas of authenticity and integrity into question. Artwork can be devalued based on the motivation of the creator. Was a piece commissioned? Was the original influence used with permissions and given proper recognition? Was the process influenced by a desire for money, fame, or social status? These abstract questions assist in the overall reception of art regardless of the quality of the final product. The intrinsic and extrinsic motivators must be evaluated with the rise of AI Technologies in art. As the programs themselves lack the ability to speak and act independently- the rationale of the programmers are what ultimately is called into question to consider moral and ethical ramifications of granting AI the credit of intellectual authorship.
As AI art continues to leak into modern markets, and collectors interested in embracing emerging technologies create demand, the schools of thought opposing and supporting the rise will continue to clash. As humans travail to define and regulate the field, the works of AI artists will continue raising questions about the nature of art and the role of true human creativity in future societies.
Check back next week to hear the affirmative side of the AI Art debate featuring a Q&A with artist and writer Joseph Munisteri who’s recent NFT collection features nine AI artworks.