My stance
I am not completely against AI as a tool but I still strive to use it as little as possible.
For pedagogical texts and blogs, I feel it is honest and preferable to struggle through words and sentences like a horse on a bike. It is more human.
Whenever I read a blog or browse open source projects, I check for AGENTS.md, and other tells in the .gitignore, and commit messages.
I am suspicious of vibe coded projects and approach them with more caution than otherwise human made content. Maybe this sentiment will change in 5-10 years time but that's not the point of this document, really.
I am not looking for a debate on this matter, I think the internet has more than enough for you if that's what you wish to engage in.
For others, like myself, I include this document to be transparent on what is human made and what AI has touched.
I wish more projects had an AI disclaimer.
Zola theme refactor
When I started this blog I had a very bare bones layout. Design, colour and theming are not my forte. After I had written quite a lot of nonsense I wanted to use a theme instead.
I tried to wrangle themes into the structure but couldn't get it to work correctly.
I got an AI agent to assist me in rewriting the structure and that helped me make some headway. I'd say the endeavour was 70-80% AI and the rest human fixing.
Grammar and typos
I write everything myself. No sentence has been AI generated.
English is not my native language, but I think it's safe for me to think of it as my clumsy second language. In some moments of self doubt I've had AI review my text and asked it something along the lines of
Review this text for grammar mistakes and typos.
Do not alter the voice of the author.and then fixed it myself if I agree. For comparison and fairness I've opted to not review this document for grammar mistakes and typos.
For a little while, a series heading had an AI generated description after the Zola refactor which slipped my notice. I replaced it with my own writing when I wrote this document.
Researching
Uiua
I do not use AI for researching Uiua. I don't think there's enough material available to be devoured by the LLMs anyway.
What I do is read over the documentation again, search through the Discord, asking my go-to nerd or asking a question in the learning channel of the Discord.
What about the Euler problems
I like struggling. I believe that learning requires struggle, at least for people as rat brained as I am. I am lucky to have a very smart group of friends that I can ask about maths and computer science whenever the fancy takes me.
I don't get the same joy researching with AI as I do with a discussion. So, for this blog I have not used AI for research.
Also, some Euler problems give you a solution guide upon completing it. If I barely make it through with brute-forcing, I can read that guide as well as the problem's discussion thread.
Finally, I really like this statement from project Euler's GenAI policy
But if the genAI gives you the answer, you have gained nothing more than a badge of shame as you submit the answer to a problem you never solved.
What if you're wrong?
I am human. I make mistakes. You human? You see mistake?
I encourage and welcome all discussion and correspondence. I usually include emails for that purpose in the section introductions. Feel free to send me corrections or feedback. I welcome it :)