This April OpenAI released Dall-E, an art AI with the potential of making artists obsolete during one of the worst economic situations the United States has faced since the recession. Many more art AI models from different companies/groups have come out since then.

Furthermore, there are many models in development that have the ability to generate 3D models and image continuity, making AI-generated animation possible.

Many technological revolutions in the past have benefited society as a whole, such as the printing press. Although these technological innovations are important when it provides a net benefit to that specific industry and society as a whole, we believe AI art programs (based on neural networks) provide no benefit whatsoever to both those working in the art field and for society as a whole, for the following reasons:

• The generated artwork devalues all other work it replicates; the artworks produced by these machine learning models can be made quicker than a human artist and potentially with the same quality, but for as little as $0. Furthermore, these generated pictures can be indistinguishable from human-made art, destroying any potential for human-made art to hold more value. Any creative endeavor, even if it is made with art AI, will thus be unsustainable financially, and an industry once worth millions will now be unable to pull in any revenue.
• Generated photos (and eventually videos), in the beginning, could lead to the misuse of generated content, especially in legal situations. Later on, photo and video evidence as a whole cannot be used as evidence whatsoever, and generated content may compromise all trust people have in any content they see.

Furthermore, the "benefits" many of the supporters of these models push are not realistically beneficial:

• "AI can improve the workflow of working artists"
  • Although they do improve the speed of art production, they do so by replacing the workflow rather than acting as a tool in the process.
• "AI can reduce artist overwork and help provide a healthier work-life balance"
  • When less work is required to complete a commodity, the commodity is worth less, and the worker is forced to work even harder to make up the difference in value. This will either lead to working the same amount, being forced to work more, or be forced out of this workforce entirely.
• "Human-made art/traditional art will rise in value"
  • Generated pictures can be indistinguishable from human-made art, and therefore cannot be trusted as actually human-made. Moreover, many mediums depend on technology and digital mediums to be distributed; comics and films need technology in order to be distributed (at the very least, comics need to be printed, and movies need to be transferred to a physical medium, both of which require technology to be done). Any medium that requires technology to function will lose any credibility as human-made, for they could have been easily mistaken as ai-generated, therefore their value will match the value of ai-generated art: $0.

Concerning the "freedoms" and "democratization" of art provided by AI art models:

• While AI art does make it easier for the average person to create artworks, it does so by taking away the freedoms of current artists. Many artists need to spend a significant amount of their life and time to express their freedom as an artist, to the point where they must depend on their art to support them. With AI art dropping the value of art, these artists lose their artistic freedom of time granted by the monetary support from their work.
• Everyone already had the freedom to make art, the only limit being the free time they had and the expectations they set up for themselves, both of which being issues that professional artists already face.
• Individuals from low income areas still require a functioning computer to use these AI models. Although art supplies may be expensive, they are no more expensive, if not drastically less expensive, then having to purchase a computer setup/phone/tablet.
• Although some disabled individuals may not have the ability to create artwork at all, AI models will just take away the value and freedoms current disabled artists have.

Additionally, many art AI models, such as Dall-e 2 and Craiyon, have also been training on data that does not belong to them. Much of this data is used without the permission of the original creators of said data, in turn violating the creators’ rights to how their own content is used. 

We are calling for the United States Government to regulate AI programs that utilize neural networks. This includes, but is not limited to, Dall-e 2, Craiyon, Imagen, Parti, DeepAI, Stable Diffusion, and GLIDE.