Moderated by Martha Brockenbrough
Umair Kazi: Director of Policy & Advocacy at the Authors Guild
This panel was full of SO MUCH information that I will be revisiting it myself but here are some highlights and great thoughts about AI.
On Definitions. Generative vs non-generative. What's the difference and where is the line?
Umair: It's true that we can distinguish the two on a tool basis vs generative AI but AI is in everything now, though it has always been around in a way. For example, spellcheck is a form of AI. It is coded, not generative. Still a form of AI. But now, when we use grammar tools, you now get a revised version of what you wrote.
The copyright office has a definition. Generative AI technology is technology that generates text, images, visual and audio material that was created by a human aka copyrightable material. Be mindful that generative AI is built on that copyrighted work. There is a line between it being a functional tool vs using it to create copyrightable content.
In using AI: Is there a moral issue or are their practical business considerations?
Umair: There are some legal boundaries but when we think about AI and authorship, you often find that no output of an generative AI is copyrightable. Reason being that copyright requires human authorship.
Courts have been working on this for many years prior to the boom of AI. For example, the "Monkey Selfie Case" where a monkey took pictures on a camera left in the forest and because their wasn't any sufficient human control over it, it brought in questionability to the copyright of the images and became problematic from a copyrightable point of view.
Martha once asked a question on Google search and Google Gemini, what appears at the top, produced the wrong answer. Martha knew this was incorrect and argued with Google Gemini but did not take the corrections which presents potential problems.
On Licensing with AI:
Umair: In training AI, the nest practice is you have to get permission from the author. What do you get for this? They are starting to look at AI rights as subsidiary rights. In model clauses, they are pushing for a recommended 80 to 90% of profits from work that is used with AI.
There are illustrative possibilities for the future, like e-books with an AI component but again, the artist should be compensated.
Martha: "AI is creating mediocre work! We don't want mediocre work!"
Umair: People say that "human written work has mistakes" but the language around imperfections should be considered. There is beauty in the humanness of art in all its imperfections.
Martha: Shakespeare and Dr. Seuss created words that have been added to the lexicon. Words will be fossilized if we continue to use AI. Example: we created Tang as powdered orange juice but it isn't as nutritious as OJ. Do we want to give our kids the literary equivalent of TANG or are they entitled to freshly squeeze HUMAN SOUL?
Grammarly? Is that okay?
Is it changing your voice? If AI is changing your voice to the point that it isn't your writing, is it worth it? Is it really making your expresson better? To judge if it's okay or not, depends on what it is doing, but now it can change your whole paragraph.
"Don't surrender your thinking for AI." - Martha
Umair: I hate to put another obligation on the artist where you have to keep track of all of your process because someone might question your authorship but idea development is a tricky one in some ways. You can use AI to ideate but you can't use it in your writing. There has to be enough authorship" in your personal work. Idea vs expression, that is the dynamic at play. Idea re not protected by copyright but the expression is. In terms of clauses, we want to create a process if your work is rejected because of AI usage where the process record can come in to help.
"I feel AI is an okay place to start (for research), but is never where you're going to end. A mama bird gets a worm, chews it up and gives it to the baby bird. That is what Google AI is kind of like." - Martha
Do you open up your work to scrapping by posting your stuff online?
Unfortunately, anything you post online is up for scrapping. We must assume that anything online, like a Google Doc, is making our stuff available. Sometimes you can opt out but some of that has changed. We need a massive political change and must assert control as a democracy over AI.
Glaze is an option for illustrators to protect their work.
Best advice for creator moving forward., Best practices and how do we cope?
Umair: Reminds me of a really wonderful quote; "the imperfect is our paradise." Though that quote is a critique of fascism, it can apply here.
Have faith because one of the things I do see is that people are not drinking the "AI Koolaid." It can be tempting to use due to insecurity but I think a lot of AI use can compromise the author and AI is not a better writer than you!
Umair Kazi is the Director of Policy & Advocacy at the Authors Guild, where he works on copyright, collective bargaining, antitrust, and artificial intelligence issues. Previously, Umair served as the Guild’s Staff Attorney, and advised members on legal issues related to the writing profession. Umair holds a J.D. from the University of Iowa College of Law and an M.F.A. in Creative Writing from Columbia University School of the Arts. He translates poetry from Urdu, with work published in several literary journals.
