AI Policy

I do not use large language models (LLMs) or any similar form of “generative AI” in any capacity in any of the writing, music, code, or images on this website.

I type my blog posts by hand in Neovim, using the codebook spell-checker to catch typos. Code is similarly written in Neovim using this list of language servers to catch errors, and this collection of formatting tools to achieve consistent use of line breaks, indentation, etc. — both categories use traditional code parsing, rather than LLMs.

For the images, I create and edit SVGs in Inkscape, touch up and scale photos in GIMP and/or ImageMagick, and cite e.g., digital signal processing diagrams from papers or other discussions when relevant. I take photos on a Pixel 8a running GrapheneOS, and the only on-device image processing is via the default hardware settings and the GrapheneOS camera app.

I keep a list of my primary music software here. While some of these programs ship with features the manufacturers describe as “AI” (e.g., Logic Pro has the ability to extract instrumental or vocal “stems” from a completed recording), I do not use these features in any of my composition or production.

I maintain a page of readings critical of “generative AI”, sorted by category, and which I regularly update. These readings are a good summary of some of my main issues with the technology.