While AI systems like DALL-E and Midjourney can generate images from text descriptions, they do not create art in the same way humans do. AI art generators are tools that produce variations of existing images based on the data they are trained on, but they lack emotional expression and personal experiences that are fundamental to human art. 

We quite liked the article in the Harvard Gazette by Liz Mineo from August 15^th, 2023: “If it wasn’t created by a human artist, is it still art?” It summaries engagements with AIs by writers like Daphne Kalotay, professor for Creative Writing at Harvard Extension School, musicians like saxophonist and composer Yosvany Terry, director of Jazz Bands, animator Ruth Stella Lingford, senior lecturer on Art, Film, and Visual Studies, and mixed-media artist Matt Saunders, professor and director of undergraduate studies in Harvard, Department of Art, Film and Visual Studies, and architect Mishe Safdie, design critic and urban planner at Graduate School of Design.

These creators emphasise that an artist can truly collaborate with AI if he or she is capable to ask the right questions. AI might be a superb mimic and quick learner, but it lacks genuine insight. In jazz, the sense of interplay and the ability to react in the moment, can’t be reproduced by AI because doing this requires having the agency to use curiosity and possess your own – and unique –  musical vocabulary. 

Of course, AI has been already used to compose for films and television. But these compositions lack surprise, emotion, and even silence. 

Walter Werzowa, creator of the AI Beethoven Symphony, on another hand, believes in the limitless possibility of AI as a creator.

There are several issues regarding the copyright of AI-generated art. Current copyright law requires a human author for a work to be eligible for protection. This leaves AI art in a legal grey area.

AI art platforms and individuals using them could potentially be liable for indirect copyright infringement by enabling and distributing content that infringes on copyrighted works used in the training data. The extent to which AI art transforms source materials versus closely imitating them will impact such determinations.

If AI art is deemed copyrightable, there are questions about who owns the rights: the AI system developer, the prompt engineer, or the end user.

AI art generators have been trained on the work of human artists without their consent or compensation. Some artists have filed lawsuits alleging copyright infringement.

In January 2023, artists Sarah Andresen, Kelly McKernan, and Karla Ortiz filed a class action lawsuit against Midjourney, Stability AI, and DeviantArt, alleging their copyrighted artwork was used to train AI models like Stable Diffusion.

The lawsuit was later amended in November 2023 to add seven more artist plaintiffs and defendant Runway AI. The complaint argues the AI companies compressed and stored billions of images, including the plaintiff’s copyrighted works, which they then used to create new AI-generated images that are strikingly like the artists’ styles. 

In October 2023, a judge dismissed most of the artists’ claims but allowed the artists to amend the complaint to clarify their theories.

 In May 2024, a judge made a tentative ruling allowing the artists’ direct and induced infringement claims against Midjourney, DeviantArt and Stability AI to proceed finding the artists plausibly alleged copies of their works and stored in Stable Diffusion. The AI companies have argued their use of the artwork to train AI models is fair use and does not constitute copyright infringement.

Several prominent writers also have expressed outrage over the unauthorized use of their books to train AI systems by companies like Meta/Facebook and OpenAI. 

Min Jin Lee, author of “Pachinko,” labelled the use of her books as ‘theft’: “I spent three decades of my life writing my books. The AI companies stole my work, time, and creativity. They stole my stories. They stole a part of me”.

Sarah Silverman, Richard Kadrey and Christopher Golden filed a lawsuit against Meta claiming the company violated their copyrights by using their books to train its LLaMA language model.

Douglas Preston, a novelist, was shocked when he saw ChatGPT generate a poem about his fictional character, realizing the AI must have ingested his books. He is now part of a class action lawsuit against OpenAI filed by the Authors Guild for using hundreds of books to train ChatGPT without consent. Other authors involved in the lawsuit include George R.R. Martin, Jodi Picoult, Michael Connelly, and Jonathan Franzen. The Authors Guild argues AI companies need to respect authors’ rights and have a licensing system to use copyrighted material.

In contrast, some authors like Ian Bogost believe their work being used to train AI is not a bad thing.

We believe that the future of AI in the arts is promising and that the landscape of the art industry will be altered while technologies advance. We will develop a GenAI tool for children to create their own Romy&Roby stories while localizing content and adding new characters once three Romy&Roby books are in the market and there is sufficient body of work – dataset – to train such an AI. We plan to do it as we want to increase diversity in the Romy and Roby universe. Our stories currently happen in a little town near the European Alps, and the environment is mirrored in how our characters’ identity is shaped. We would like to give Romy and Roby an opportunity to be at home in other countries, and different cultural environments!