As of this writing, AI and agents are so pervasive throughout the internet, its creators, its implementation, and its functionality that ignoring its impact (and potential misuse) would be naive and out of touch.

The purpose of buttermilk is to create things you, as a human, actually want to read. And, to ensure you can trust what's written here. Regurgitating slop only diminishes the quality of buttermilk's content and erodes the trust you place in buttermilk as a reader. Trust is slow to earn and instantaneous to corrode. The time you spend on buttermilk is precious and you should feel it is time well spent.

As a result, buttermilk adheres to the following policy on AI.

• ~/buttermilk $ cat policy/creativity.txt

  Human creativity is emphasized and artistry is credited.

  The internet was built on human creativity and buttermilk seeks to add to that creativity, not detract from it. When you read content on buttermilk, you can have confidence that it’s backed by human creation and not veiled slop. Human sources are credited wherever possible. AI should be a catalyst of creativity, not its replacement.

• ~/buttermilk $ cat policy/content.txt

  No portion of any post is written, edited, or outlined by an LLM.

  This means when you read a post on buttermilk, you know it came from a human. To be explicit:

  • No portion of any post is written by an LLM. This means the initial draft and subsequent writing was created by a human. Content from an LLM is never copy-pasted or fully generated into buttermilk posts.
  • No portion of any post is edited by an LLM. This means initial drafts and sections are not passed to LLMs to be cleaned up, edited, or reworked and then posted on buttermilk. Posts may be reviewed by LLMs during the editing process, but only as a collaborative brainstorming partner, never as a direct author or content contributor.
  • No portion of any post is outlined by an LLM. This means the outline, structure, and skeleton of posts is not generated by an LLM. LLMs may be used during research and editing, but the format and flow of the content you're reading will always be created by a human.

• ~/buttermilk $ cat policy/thesis.txt

  The thesis of buttermilk posts is defined by a human.

  The content of posts on buttermilk always represents the thoughts, opinions, and ideas of a human. A human defines the premise, core arguments, and thesis of every post. LLMs may be used during the editing process and during brainstorming of ideas, but the core thesis of posts are never copied or regurgitated from an LLM.

• ~/buttermilk $ cat policy/images.txt

  No images, music, or videos are generated by AI.

  All audiovisual content (such as images, figures, audio, and videos) on buttermilk are created by humans. If AI-generated audiovisual content is unknowingly and unintentionally included in a post, best efforts will be made to remove them as soon as the accident is recognized. Audiovisual content that is not authored by buttermilk will always include a source. Third-party audiovisual content is always licensed appropriately.

• ~/buttermilk $ cat policy/sources.txt

  LLMs are not used as trusted sources.

  LLMs may be used during the research and brainstorming phase of content creation. LLMs and agents may reference trusted sources during brainstorming, and if buttermilk content references sources identified by an LLM, they are thoroughly reviewed by a human before load-bearing inclusion in content. LLM output is never listed (or indirectly referenced) as a trusted source. Trusted sources are human reviewed for accuracy and rejected if they're noticeably AI-generated or overly AI-influenced, even if the source doesn't directly attribute its content to AI.

• ~/buttermilk $ cat policy/code.txt

  buttermilk.fyi is designed by a human and generated by an LLM.

  The structure, format, and functionality of buttermilk.fyi is designed by a human. The intention of buttermilk, such as what pages you see, how you interact with it, and specific features are all designed by a human. LLMs may be used in the design phase for brainstorming and planning. The final implementation of buttermilk.fyi is executed by an AI agent.

• ~/buttermilk $ cat policy/research.txt

  LLMs may be used during brainstorming and research.

  Ideas come from many places while building content on buttermilk, and it's fine for AI to be one of them. Sources may be identified by AI but are only referenced and trusted in buttermilk content if they have been thoroughly verified by a human.

Inspired by Complexly's AI Policy.