Given the increasingly heated online chatter, including some outright witch-hunts, that have been happening about authors and their use of Large Language Models (LLMs, often imprecisely called “AI”), and given the fact that I am relatively open about my use of LLMs in some statements, I think it best if I go ahead and issue the below formal statement about my use of such models.
Philosophy
I am exquisitely aware of, and extremely concerned about, the ethical issues surrounding use of LLMs. They are many. Studies have shown that more and more written material generated solely using LLMs is being released into the marketplace. LLM usage has also been linked to loss of critical thinking and learning among students, is known to be error-prone and sycophantic, is often of poor quality, and, perhaps worst of all, often involves the use of models that were trained using improperly and even illegally obtained data.
Let me be clear: these are not minor issues, and I do not see them that way. However, I also feel that a significant part of today’s climate and attitudes about LLMs, particularly with respect to the online and literary communities, have gone to the opposite extreme.
Whether or not we like it, LLMs are here to stay. As a working professional in a business field, I have already been integrating their use into my activities and they are invaluable. The opportunities for advanced analysis, pattern recognition, and automation of repetitive tasks are significant — and carry great impact. New uses for LLMs are being discovered every day, and the development of the technology isn’t just continuing; it’s accelerating.
LLMs and similar tools are also older and far more prevalent than many either realize, or would like to acknowledge. One of the earliest uses of this technology, before it evolved into modern LLMs, was the use of spelling and grammar checkers in word processing applications. Tools such as Automated Insights’ StatSheet and Grammarly have already existed for decades, and the first completely computer-generated book was actually published in 1984.
While it is popular and even emotionally comforting to declare yourself “completely AI-free,” it’s no longer realistic. Many tools authors already rely on — including spelling correction, grammar analysis, predictive text, and automated editorial assistance — exist on the same technological continuum as modern LLMs, even if generative systems raise additional ethical concerns.
At this point, the question is not whether an author is going to use LLMs/AI. It’s how. The other relevant question is: are we, as authors, going to continue in our pattern of denial, or are we going to actively engage so that we can be partners and participants who actively help shape future evolution of these technologies in ethical directions?
As I’ve expressed on my main author web site, I prefer the latter to the former.
Use of LLMs and Similar Models
To that end, and in the interests of full disclosure, I will confirm that yes, I use spelling and grammar checkers; LLMs such as ChatGPT, Claude, and Gemini; and several other applications that have LLMs/AI built into their underlying programming structures. However, I am careful to stay within strict guardrails with respect to that usage. My usage is explicitly informed by the Authors Guild’s best practices and I have taken several steps to ensure continued ethical usage.
I will, and do, use models for the following activities:
- Brainstorming;
- Checking structure and ensuring cohesive and realistic plots;
- Translation of words and sentences originally written in English but meant to be rendered in a different language;
- Generation or refinement of single words and brief phrases; and
- Research support.
I absolutely do not use these models for large-scale generation or editing of prose. Every sentence I publish is my own. For those applications I use most often, I have turned off the sharing of my material to the larger community. In addition, I have included explicit instructions to the applications against proactively offering me editing or creative assistance unless specifically and directly asked. When I do ask for such assistance, I always request at least three options from which I can choose and the final choice is my own.
I am confident that anything I publish under my own name and/or pen names is copyrightable — and, in fact, in most cases I actively do claim such copyright protection. The sole exception to that sentence is situations where I forego claims of copyright protection because an item is fan fiction. There are no situations where I disclaim copyright because something was created by generative LLMs/AI.
Purpose of this Statement
Even though I strictly limit my use of LLMs/AI applications, the fact that I use them at all still disqualifies me from some authors’ groups, contest entries, and other literary activities. I am aware of this, and if I become aware that my use of these applications violates a situation’s terms, I will either refrain from becoming involved in the first place, or voluntarily withdraw from such involvement.
I do not ever attempt to obfuscate or “sneak” my LLM/AI application usage into places where it is not welcomed. However, I will not discontinue such usage simply to comply with others’ guidelines or assuage others’ concerns. My choice to use them is freely and ethically made, and I insist on the right to continue making it.
Should you have any further questions about my use of LLMs and AI, please feel free to reach out and we can discuss.