More On Our "A.I." Policy
Not that we've expressly stated it on just about every page of this site, but whereas:
- Current models of generative A.I./large language models (LLMs) utilize datasets composed nearly entirely of stolen creative and other intellectual property;
- The resources required to house and operate material infrastructure for these products including space, water, and energy displace and overwhelmingly impact low-income communities in practices which are actively anti-human;
- The purveyors of this technology speak openly about their disdain for the human artists, creators, and workers whom their products seek to "replace";
- The intellectual impact of A.I. use on an individual basis shows a decline in cognitive ability and critical thinking as a skill;
works produced in whole or in part by generative A.I./LLM services and products are not welcome at Sound & Color. Generative A.I. is a product or service utilized to create "original" content based on a prompt or request, such as Claude, Midjourney, or ChatGPT. We will go pretty scorched-earth about violations of this policy. Your perceived need to integrate these products as part of your process is your business, but this is a boundary for us and we will require that you respect it.
We do acknowledge that there are accessibility applications for certain technologies billed as "A.I.", hence our emphasis on the generative piece. Acceptable use of these technologies include:
- screen-reading
- spellcheck
- transcription or note-taking
A.I. used for translation is not allowed; get a human translator.
But how will you verify?
At this time, there is no perfect system for detecting A.I.-derived stories. The technology being fed all of human creativity reproduces an amalgamation of all of it and then compares all new works to itself as an author. Things like stylistic tics, vocabulary choices, rhythm, and turns of phrase can be naturally occurring based on the storytelling traditions from anywhere in the world. We're not in a position to tell you that the way your people write reads like A.I. And there's no way to trust that running work through an A.I. checker doesn't also feed its dataset to train and refine it, forcing litigation over the dishonest practice a decade later for a settlement to be paid in peanuts.
On the other side, asking for "proof" of human authorship by requiring submission of working documents and hand-drafted notes feels almost sacrilegious.
So.
When you submit your work, you will of course be required to check a box certifying that the story complies with our policies. All submissions will work their way through our first readers and executive editor, being reviewed by no fewer than 3 experienced, story-loving humans.
If we love your story and select it for publication, congratulations! We will notify you by email and invite you to an editorial (video) call before a contract is issued. We want to get to know our authors and hope this is an exciting opportunity especially for new authors to talk about their work with people who already love it, and to have a sense of what our vision is for the editorial process. It also makes it easier for us to defend against external accusations of works being A.I. generated in the future.
That's it! Your verification is a conversation in which you are required to be a person until and unless an ethical and accurate alternative standard of authentication is developed. We will also require the presentation of evidence supporting any accusations of policy violations, just as we would for accusations of plagiarism.
We envision the author-editor relationship as one of mutual respect. It has always relied on trust and enthusiasm for the art of storytelling. We're excited to work with authors who share that vision.
Questions? Contact Us.
