Here is how I’ve used generative AI/LLM tools, why, and why I stopped. I’ve also included my personal answers to some common arguments for the use of generative AI platforms like ChatGPT, Claude, etc. I have never used generative AI tools to replace “my voice” in my writing or my artwork. I find it far more enjoyable to actually “do” the things I do rather than to have them done for me.

Previous use

In the past, I used ChatGPT to generate example programming functions, to learn from. I adapted those examples into my own code. This was typical for personal projects like my installation of Obsidian (see my “Links and Tools page”). I typically used it as a simple learning tool while understanding it’s limits and flaws. I asked it questions like “can you explain this REGEX snippet to me if I paste it here?”, and then I learned something from the exchange and carried that forward. I learned some dataview javascript that I expanded upon and implemented into my Obsidian vaults. I now have a stronger grasp on writing this kind of code without assistance or now understand better how to use the documentation and its included examples provided for dataview (and it’s various flavors) to extrapolate scripts to accomplish my goals.

I used a local installation of Stable Diffusion for quickly iterating ideas on color, composition, and as references, not ever to make finished or near-finished works which I intended to represent my work as an artist. I have followed this same workflow previously in Photoshop using stock media and in physical form long ago by cutting up magazines to iterate color, composition, and references. For me, it was a tool in the workflow that helped ideate; I never used it to create anything further in the process than references. Like I said above, I want to paint and draw the things myself because that’s what makes one an artist.

It was never my intent to use any of these tools in a manner designed to harm, deceive, or replace anyone. I stopped using all generative AI in early 2025.

Why I stopped

Now that I’ve seen how the corporate system developing and implementing these technologies has no intention of respecting human creativity or IP rights, I’ve chosen to abandon even the limited use I outlined above. AI companies and those heavily integrating and pushing it have shown their priorities and have at every opportunity chosen money before humanity and environment. I’ve also abandoned it because many peers in the arts community have become almost militantly opposed to such technology. I share their concerns, and have made the change, but I can’t stand their sanctimonious bullshit piety about it.

I went further

Once I’d made the decision to abandon generative ai companies, websites, and apps, the next logical step for me was to figure out how to also remove it from other workflows in my day-to-day. Given the rapid pervasive pandemic-like infection of AI in everything, I accept that I won’t be perfect in my attempts. My like I feel that 1,000 people doing what they can to help the environment without being shamed, is better than 1 person doing it perfectly.

So, I paid attention to where else in my life AI was creeping in, and figured out if and how I could prevent it. This has led to me cancelling subscriptions for Microsoft Office, Adobe Creative Cloud, and others. I’ve uninstalled their apps and found alternatives that haven’t–so far–implemented generative AI in ways that I can’t turn off. Firefox v148 introduced a great feature that allows me to turn off all AI within their browser with just a single click. I would love for more developers to implement such a solution! Even before they offered that, I researched and implemented ways to adjust hidden settings to turn off a lot of AI features in that browser. I also use a program call “ShutUp10++“, which allows me to not only turn off much of the AI garbage in Windows, but most, if not all of their data harvesting/telemetry bullshit as well. Look for more resource lists coming up related to this topic. I will link them here when available.

Applied sciences usage

I’ve made a distinction, for myself, between generative AI and what I’m calling “applied sciences use”. I support this kind of use, despite it also have the environmental draw-backs that is baked into the technology, because it is genuinely helping humanity be more effective at scientific research, discoveries, and innovations. I’ve been figuring out the best way to explain this type of use, and here’s what I’ve come up with. It analyses and processes inputs using a narrowly defined outcome, is built from data related to accomplishing that outcome that was ethically sourced, in which the outcome forwards a scientifically sound purpose. For example, astrophysicists used a library of hundreds of thousands of photographs of portions of the night sky and galaxies to train an AI model to classify photographs of galaxies into one of several categories identifying it’s shape type (spiral, elliptical, irregular, etc.). The work had been performed by scientists, and groups of trained volunteers called “citizen scientists”, but with new platforms like the James Webb Space Telescope (JWST) and Euclid Space Telescope (EST) coming online the number of new photographs that need classification has surpassed human capacity within hiring/funding limits. The model does the work much faster, and has at least as good an accuracy as the citizen scientists. It flags ones it’s unsure of for human help.