On the Use of LLMs and AI vs. Writers

 

Today, I want to share some thoughts on AI and the use of Large Language Models (LLMs). Back in April, I presented my perspective on AI in an article titled “The AI Factor.” In this piece, I’m diving deeper into the subject, this time based on a specific experience I had recently.

Before I get into the details, let me share a bit about myself. I’m not one to follow trends blindly. I did that during my early high school years, but I eventually grew out of it and started thinking independently—around the same time I began writing. This means that just because many people hold strong opinions on a topic, it doesn’t necessarily mean I agree with them. I respect their right to their opinions, regardless of how they’ve formed them. Now that you have this background, let’s dive into the matter at hand.

Last Friday, I received an email from my self-publishing platform inviting me to take a survey about LLMs and the creation of original content by writers. They were exploring new revenue opportunities and wanted feedback. Recognizing that this is a sensitive subject, I decided to participate. The survey asked three questions, though I wish I’d written them down as I can’t recall the exact wording.

The first question was about the general use of LLMs, which I’ve already discussed in my previous article. In short, I’m comfortable with using AI as a tool in my writing process. To me, AI serves as a helpful aid for brainstorming, editing, improving grammar, and even providing inspiration when I’m stuck. It acts as a partner or assistant, not a replacement. My creativity remains at the forefront, with AI serving as support. This is my baseline perspective.

The second question was more complex, involving an example scenario. It asked how I’d feel about using my written work to train LLMs, which would then be used to train 200,000 employees at a large company. I was also asked to select an amount I’d accept as fair compensation from a list provided. I found myself quite comfortable with this scenario and chose a relatively small amount as fair compensation. I value education and training, and when I considered my work being used to train AI for internal, non-commercial educational purposes, I didn’t see it as a significant issue.

The third question presented a similar scenario, but with a critical difference: instead of training employees, the LLMs would be used to allow customers to create personalized stories in the style of their favorite authors. This added a commercial and competitive element to the scenario. In this case, I felt much more hesitant. While I was comfortable with the idea of my work being used for training employees, the notion of my style being mimicked for commercial purposes raised concerns. I selected the highest compensation rate available, though I did consider not accepting it at all.

This discrepancy in my responses made me reflect on what made these two situations so different in my mind. The second scenario raised significant concerns about intellectual property. The idea of AI mimicking an author’s unique style introduces ethical and economic questions. If consumers can generate personalized stories in my writing style, it could devalue my original work and create confusion in the marketplace.

As writers and authors, we highly value the integrity of our creative voices. It’s what makes us unique, regardless of our skill level. Allowing LLMs to mimic specific authors broadens the discussion to include questions of authorship and ownership, such as:

• Who should own the rights to AI-generated content that mimics a specific author’s style? Should the original author have a say in how their style is used or commercialized?
• How do we determine what constitutes fair compensation when an author’s work is used to train an AI?
• Should there be ongoing royalties, or is a one-time fee sufficient?
• What happens to the value of creativity if AI can mimic human authors? Does it diminish the uniqueness of our individual voices, or does it open up new possibilities for creativity?

I don’t yet have the answers to these questions. However, the fact that I didn’t outright reject the idea of my work being used to train LLMs tells me that I’m willing to consider different possibilities and think them over carefully.