AI Disclosure

Last Updated: 18 June 2026
The full history of changes to this page can be tracked here.

I have noticed an increase in honest online content creators creating "AI Disclosure" pages on their websites or social media accounts. I think this is a great trend, and would very much like to participate in disclosing to anyone who consumes the content I put out, exactly what it is that they are consuming. This page attempts to make clear (a) What I do and do not currently use AI for, (b) what I have used AI for in the past, (c) how I intend to use AI going forward, and (d) my rational. I have tried to be as thorough as possible, but as always, I'm happy to answer any questions that might not have been addressed here.

What I Don't Use

It is currently my policy to not use any AI agent, LLM or any other form of Artificial Intelligence to generate any content that I make public, unless otherwise noted. It is also my intention to maintain this policy essentially forever. My objection to AI is two fold, and relates not only to the usefulness or the tools, but also the ethics of using the tools. Both of these objections would need to be remedied for me to consider changing my policy.

What I Do Use

While it is my policy not to use AI to generate content that I sign with my name, or run on my systems, I have no way of knowing which developers and online creators share this policy. I am aware of a few projects that attempt to make it easier to fully block oneself from using any software that was created with the assistance of AI, however I view this effort as essentially DOA. At this point even tools as fundamental as the Linux kernel allow for AI-assisted contributions. Since I do not have the expertise or time to created my own operating system kernel from scratch, I have accepted that much of the software and tools that I might use on a daily basis may have been developed with the assistance of AI.

Additionally there are a few dependencies that are crucial to the running of this website. Many of these, do use AI in some form or another. I do my best to keep a pretty close eye on all of these tools, and have a back-up ready to swap in when/if any of these tools should cross a line that I view as being too far. A full list of these tools and links to their relevant documentation can be found on my website's Terms of Service & Privacy Policy page, but I have made some quick links available in the table below:

What I Have Used in the Past

For just about everything I have ever put out on my website, GitHub or YouTube channel, I have had a policy of not using LLM's, AI Agents or any other form of Artificial Intelligence to create or enhance my work. There have however been a few examples in which I have broken my own rules, and used a LLM to generate content:

• The thumbnail for my YouTube video: "The Best Value in the History of Literature" was generated with OpenAI's GPT-4o image generation. My original intention was to use the AI generated image as a reference to build my own, better thumbnail in Gimp. But I got annoyed with using Gimp, because Gimp is terrible, and instead just took the lazy route, added some title text to the thumbnail and hit upload. I'm now refusing to change the thumbnail so I'm reminded of how stupid of an idea this was every-time I scroll through my YouTube videos.
• A ground-up rebuild of this site that took place in 2025 originally used 11ty as a static site generator and Airtable as an easy CMS-ish solution. When I was figuring out how to make these two tools work together, I used ChatGPT to fill in the gaps of my JavaScript knowledge. Initially, AI generated solutions were added with commit 16a03f5, and they have been slowly worked out as I've found more straightforward ways to accomplish what I was trying to do. All AI contributions were fully removed in September of 2026 with commit 88a8616 when I stopped using Airtable as a back-end, and replaced it with a JSON table.
• I decided that I wanted to add descriptions to the various categories of images in my Wallpapers repository on GitHub. I used ChatGPT to generate a description of the categories, and left it in readme.md for several months. The AI generated text was added with commit 4ea0c60 and removed with commit 308699b.
• I have used Grammarly in the past to edit several of my essays. Most of my Grammarly usage likely took place before it become an "AI Partner", as I very rarely can be bothered to paste an essay from Neovim into a browser before committing it, but it's reasonable to assume that there may have been some over-lap and at least some of my grammar may have been corrected with an AI agent.

Philosophy

1. I view AI, particularly LLM's and the current iteration of large-scale AI development to be fundamentally flawed in terms of trustworthy output. There are any number of articles to be found about AI hallucinations or LLM's convincing people to kill themselve's that prove this much. I also wrote a longer form, somewhat humorous essay that explains my initial rational for halting my own usage of AI and LLMs.
2. Even if these tools were trustworthy, and did not require constant human baby-sitting to be even marginally useful, I find the ethics of using said tools to be incredibly dubious. Whether your specific concern is related to AI weapons, AI surveillance, AI generated pornography andior AI girlfriends and boyfriends, AI data center's power and water usage (and the local government corruption that it always seems to create), AI art, autonomous driving or the less than trustworthy "titans" of industry who are currently in charge of the world's biggest AI companies - there are more than enough ethical concerns to dissuade one from using these kind of tools for the foreseeable future.
3. Additionally I view AI as fundamentally anti-human in it's goals. Tools like the cotton-gen, car and train have all aimed to replace human labor in the past, and come with their various economic upsets and controversies. However, to my knowledge there has never been a tool that has aimed to replace human intelligence and ingenuity before. The pronounced goal of AGI is precisely this, and while I'm incredibly skeptical of the eventual reality of AGI, as a human being I don't think I could sign on to support something that aims to replace me and my fellow humans.

Local AI Models

It seems worth mentioning that many of the problems that stop me from using industrialized language models, are not present in the use of local AI models running on one's own hardware. I've yet to bring up the wide-spread copyright infringement or data collection that AI companies engage in thus far on this page; but these problems are mitigated to some extent by running a local AI model, along with some of my other concerns (Example: the water and power usage mentioned earlier).

To be clear, as of the writing of this article (a) I have yet to actually set up any local AI solution on my own hardware, and (b) local AI models do not solve all of the issues present within the field of AI. Local AI models are just as likely to hallucinate and are likely to be just as inaccurate in all the ways that industrialized AI models are. Of course this does not even mention my most salient point from earlier: the anti-human aims of AI and particularly the AGI that we're being promised by heads of companies who have a vested interest in lying to us about the timeline andior possibility of AGI.

That being said, it seems worth mentioning that I am not uninterested in the possibilities of locally managed AI agents and LLMs. The ability to train these AIs on a data set of one's choosing and assurance that my use of a local AI agent isn't taking someone's job if I control the AI are great points in favor of this form of the technology. However there are still enough points against the technology to have prevented me from setting up my own AI models locally thus far; though some of this might also be due to the rapidly increasing storage, GPU and memory prices preventing me from building an AI rig of my own. In any case, it seems unlikely that any application I would have for local AI would find it's way into the pipeline that I use to publish material online, so it's probably a moot point anyways.