Yes, I use Artificial Intelligence in my writings.

And yes, I use generative AI. Not for everything.

How I Use AI in My Writing

As for how I use it - that varies based on my project, mood, energy (and dopamine levels, etc.). Not every piece I publish will use everything listed below. Heck, most of them won't. That said, I'm keeping the disclaimer to cover my ass in the rare occasion that I just want to get past a mental roadblock.

Fleshing out ideas: I prompt my AI writing partner to act as a critic and approach me using the Socratic method. I also tell it to bring receipts - quotations, articles, research papers (with links) to challenge me on the stance I want to take. I ask it to take on the role of naysayers and point out the weakest links in my arguments. I then apply some of the feedback given. What this means is that I decide what I'm writing about - I just tighten up what I want to get across.

Research: I ask my AI writing partner to look up information for me and to not only share the explanation but also the websites it used in its research. I then visit those websites to find supporting information. Is the author a credible source (and if so, how), and are other sites/works cited to back up what was said (erm... written)?

Outlining: At this point I know what I want to write about, my supporting points, and what outside views I want to bring in. However, that blank screen is an annoyance, so I ask my AI writing partner to give me a list of writing prompts for me to use when fleshing out my first draft.

Revision/Tone/Flow: Once the first draft is to a "I feel done" point, I do initial proofreading myself. My goal is to make what I write sound like how I talk (with sidebars because, let's face it, that's how a neurodivergent brain operates). Once I complete that, I run it past my AI writing partner for feedback on grammar, flow, and tone. And then I use maybe 10% of the feedback. What I'm looking for are major gaffes that I overlooked, or phrases that may make sense to me because of how my brain is wired, but may not make sense to the average person. One of my tools focuses on readability: making sure the sentences aren't too complex, and that I'm at the 6th-8th grade level.

What About Image Generation?

I use image generation to create graphics for my AI Roleplay chatbots. That is standard practice in the hobby (the vast majority of us don't make money in this niche; hence my calling it a hobby).

I feel less inclined to use AI-generated images for my overall branding (website cover images, featured images for posts, etc.) because purely text-based image generation doesn't capture my style, and I don't have the mental spell slots to do a deep dive into ControlNet workflows.

So what do I do here? I jump on Canva, grab a bunch of stock photos, and mash them up. I've been doing graphics on the computer since the late 90's, when we didn't have touchpads, and I had to make brush strokes with mouse clicks. I understand color theory and basic design elements. The AI used here is assistive - color matching, auto-correcting tone, removing backgrounds, etc.

The Tools I Use

PlotDrive: This is a token-based subscription platform that acts like a cowriter/copilot. The idea here is to not have it do everything. I'm still in charge of the overall workflow and the final written piece. It offers guidance and suggestions, but I still call the shots. I can access all AI models (GPT, Claude, Gemini, and hundreds of others), each with different token costs. The reason I went with a token-based service is that having the support be finite means I have to be strategic in how I use it. This reduces the risk of tool dependency that can occur with unlimited access.

GPT/Claude/Gemini (free mode only): While my use of these has become far less frequent since adopting PlotDrive, I will still disclose them. I use them for fine-tuning ideas and preliminary research.

ComfyUI (run on my computer): An open-source graphical user interface (GUI) that uses a node-based approach to visually construct ControlNet workflows for AI image generation. Since this uses my own computer, the final quality is affected by my system (a 3-year-old gaming laptop).

SeaArt: An all-in-one platform for AI-assisted image generation. It follows a simplified version of the ControlNet workflow system. If I feel especially nerdy, I can also transition my ComfyUI workflows to this platform, at which point the subscription plan pays for the server's processors to perform the final image generation.