The Badwriter: Stories of Copyright, Redaction, and Mediocritizing in the Wind
The recent lawsuits[1] that claim that some authors’ copyrighted books were used without permission as part of the training dataset for ChatGPT and similar AI models could be disregarding the principles of fair use and transformative use in copyright law. Fair use is a doctrine that allows for limited use of copyrighted material without seeking permission from the copyright holder under certain circumstances.[5] In this case, there are a few points to consider:
Transformative Use: AI models like ChatGPT do not merely reproduce the copyrighted books but use them to create something new and transformative. The purpose of the AI models is not to replicate the original books but to generate original and creative content based on a wide range of input data. This transformation could fall under the umbrella of fair use. Art is considered fair if it is “transformative,” or adds “something new, with a further purpose or different character.”[2] If one makes an effort to try to understand what AI is, how it works, and actually stops to think about what the acronym GPT stands for, Generative Pre-training Transformer, one might understand the power of AI’s true creativity.
Nature of the Work: Copyright law often treats factual or non-fiction works differently from creative and fictional works. If the books in question are factual or non-fictional, they might receive less protection under copyright law compared to highly creative fictional works.
Amount and Substantiality: The AI models likely do not use entire books verbatim, but rather small snippets or portions as part of a vast and diverse training dataset. This limited use might not negatively impact the market value or the core essence of the original books. How much has AI changed a piece of writing to avoid copyright is subjective depending on the particular court that is hearing the case.[2]
Market Effect: It could be argued that the use of copyrighted material in AI training datasets does not serve as a substitute for the original books. The AI-generated content is not a replacement for reading the actual books, and consumers seeking the experience of reading the books would not be satisfied with AI generated content. But one factor that is taken into account is the effect of the use upon the potential market for or value of the claimed copyrighted work.[2] Do self-proclaimed “artists” fear their “job” will be taken away? Will the world be able to progress faster when those failed “artists” are given more opportunities to realize ikigai[3] in providing something they are good at and the world needs, not just because they love to do it? Generative AI has done away with mediocre artists and writers who were never part of the market anyways … because those failures produced mediocrity, something that generative AI tools can easily pump out now with the weakest of prompts.[4]
Educational and Research Purposes: AI models like ChatGPT are often used for educational, research, and non-commercial purposes, which align with the fair use doctrine. This ultimately makes education more democratic.[6] The transformative nature of AI-generated content contributes to these purposes without significantly harming the potential market for the original books.
It’s important to note that copyright law can be complex and interpretations can vary. The legal outcome of such a lawsuit would depend on a variety of factors, including specific jurisdiction, the nature of the copyrighted material, and the transformative nature of the AI models’ use of the material.
#ai #ChatGPT #copyright
[2] https://www.epgdlaw.com/how-can-my-artwork-steer-clear-of-copyright-infringement/
[3] https://miro.medium.com/v2/resize:fit:1200/1*nrsHlAElSt87sjfzqkIRzw.jpeg
[4] https://medium.com/@pshanosky/for-now-all-that-ai-will-kill-is-the-mediocre-writer-6842f2bc3e84
