Apparently, Chat GPT is going to offer exceptionally explicit romantic content in the happy New Year. I would write the name of the romance genre, but my prudish host (what a laugh!) insists on substituting the word "Bible" for the word I want to use. Ellora's Cave was famous for it.... the literary genre, not for scripture. Amazon now sequesters and even de-platforms this type (heavy romance) of novel or novella.
https://www.bbc.com/news/articles/cpd2qv58yl5o
Now, I wonder where Chat GTP got their source material? I wonder whether any respectable authors' association or guild or union will have the courage to sue. I doubt it.
Legal blogger Aaron K Nodolf of Michael Best wrote a fine blog about the difference between input and output.
https://insights.michaelbest.com/post/102lq6q/fair-use-may-shield-llm-training-using-copyrighted-materials-but-output-infringe#page=1
As one of the authors with a potential claim in the Anthropic class action suit, I don't agree with Aaron K. Nodolf's premise that the training materials were legally obtained, and therefore the use was "fair use". My contracts with publishers were always limited, and never gave them the rights to do what Mr. Nodolf assumes they had.
His more interesting assertion is
"There are numerous authors, publishers, artists, and more who have had their materials used for such efforts, [legally obtained] and challenging the training coalesces these groups to fight these training efforts. One benefit, if these groups prevail, is that any output likely would not include their copyrighted material because the LLMs would not be trained on that material."
Does that mean that "output" such as Chat GPT may use for its "Bible-not" offerings will be based on written words that were not copyrighted?
Surely, with the launch only a few months away, how is that going to be possible?
Another random thought is that "Educational" and "Not Educational" were different categories for the online Anthropic case. Since I don't write Educational copyrighted works, I don't know anything about it, but, I wonder how Chat (or its rivals) developed the useful factual articles. Are the facts scraped from live authors' writings? What about the service whereby teachers, tutors and professors can scan works submitted by students to find out if the student's essay is plagiarized. Shakespeare is not going to mind, but what about Orson Scott Card? Or Ronald B Tobias (20 Master Plots And How To Build Them)?
Mr, Nodolf also suggests that copyright owners will have to sue individually over infringing output. I wonder if that is true or possible, except for in cases of unauthorized depictions of cartoon characters. It is expensive to register a copyright, and even more expensive to sue an alleged infringer, even since improvements in copyright law.
I know that I rant a lot about what AI does to freedom of speech, freedom of thought, literacy and grammar, but if one cannot write "Bible" (not the word that I typed!) and uncontrollable Autocorrect changes "content in" to "contention", we writers are not safe. We cannot write what we want to write. We may be told that our productivity is improved when AI anticipates what we want to write, but, I should think that the proofreading time is tripled because of all the mistakes that AI introduces.
All the best,
Rowena Cherry SPACE SNARK™
Posts
No comments:
Post a Comment