TL;DR
Where LLM is NOT used
- For generating text on this website. Every letter is written and typed by me (a human).
- I do not take
/s/before/after/gkind of suggestions. I only ask for abstract feedback (except, of course grammatical errors).
Where LLM is used
- For maintenance of this website, fixing bugs, redesigns etc.
- For research related to some topic, gathering supporting data, sources etc.
- For getting a review on structure, flow of the article and catching grammatical errors.
Motivation
Since the recent popularity of LLMs, there has been a wave of LLM-generated slop on the internet, and I don't like consuming such content, I don't like meatproxyies, and I believe most people don't like them, too. I saw a lot of other blogs that I enjoy reading (mainly from hackernews or lobsters) have similar declarations, so I thought I would add a page describing how I use AI/LLM for this blog. I also realised that some people may have very strong opinions (like Codeberg's policy to ban repos that use AI generated code), and so, me being transparent about how I use AI/LLMs in my writing can help those people to decide quickly if they don't agree with my opinions on "Fair use of AI". They can skip engaging with the content. A link to this policy is added to all the blog posts, at the top.
Where LLMs are used
I use AI in my daily life, for researching about a particular topic, for gathering supporting data, etc, much like how I used search engines before the AI era. I do not blindly add all those things to the blogs. Each source or data I get from AI is validated and reviewed before using it (Yeah, I know it seems obvious. But trust me, it is not, for a lot of people). I use AI agents for maintenance of this website, that includes adding some functionality that I want, some bug fixes, or maybe some redesign, the code for this website is opensource, and is available on github.
I do use LLMs for getting abstract feedback on the overall quality, structure and flow of the piece. After the feedback, I edit
and modify it myself. I never use the replace ... with ... type of feedback, in fact, I make it a point to actively avoid the exact thing
it suggested at all costs. That means the "AI Voice" that we all are so tired of, never seeps into the content. And I feel this is fair use.
I am an engineer, not a writer, and English is not my first language. I am not as articulate as I'd like to be, but it is what it is.
Maybe, after some time I will be, and won't need any help from an LLM, at that point I may skip the review part.
Where LLMs are NOT used
Apart from the things listed above, LLMs are not used anywhere, all the content is thought out and typed by a human. Thoughts and opinions are my own, and not those of any LLM.