AI Policy

The following is my personal policy for AI usage. It governs how I permit myself to use large language models and generative AI. I do not expect others to abide by these principles; it is here so I can hold myself accountable and point to it when people ask how I use AI.

As generative AI technologies are constantly evolving, I may change this policy in the future.

Terminology

I avoid using the general term "AI" when a more precise term is appropriate. I prefer these terms:

• "Generative AI" refers to the class of machine learning models that produce images, video, audio, text, or other media content.
• "Large language model" (LLM) describes a specific type of generative AI that receives text as an input and generates text as a response.

Principles

I do not provide my own personal information or the personal information of others to LLMs. I never send LLM-produced work to others without reviewing it completely first.

I do not use LLMs to produce English writing that I later implicitly or explicitly pass off as my own. If I include LLM-produced English writing in a written work of my own, I explicitly mark it as LLM-produced. When writing in languages other than English, I may use LLMs to review drafts so that I may reduce errors and improve clarity and idiomaticity.

I sometimes use LLMs to write code. I do not ask others to review code generated by an LLM on my behalf without thoroughly reviewing it myself first. I use LLMs to search through, understand, modify, and debug existing code.

I do not use generative AI to create image or video content. I may use generative AI image editing tools (such as Adobe Lightroom's Generative Remove) to edit content of existing images.

Recommended reading

• Writes and Write-Nots

Inspired by Sloane's AI policy and the /ai 'manifesto'