How I use AI
Dates
As of Oct 22, 2026, this is how I use generative AI tools.
Any updates to the date will happen here. :)
Code
All my code is generated using AI.
Blog posts
My blog posts are mostly written by me. When a post is partly or entirely written by AI, I will label it.
English is my second language. Even though I've been learning it for about ten years (from when I was around 6 years old), I still consider myself an intermediate writer when it comes to complex articles that require a lot of technical terminology and complex logic.
When a post covers a long debugging session, I sometimes use AI to write the entire article. I can struggle with explaining the process precisely.
By "entirely written by AI", I mean that I create an outline of what the post should cover, and AI writes the article from that outline. I then fact-check the entire article myself before publishing it.
I also use AI to catch grammar errors and remove filler without rewriting my ideas or changing my voice. I still consider that my own writing, so grammar correction alone does not get an AI writing label.
What the labels mean
- Entirely written by me: No AI was involved at all, including grammar checks.
- Partly written with AI: Some of the writing was generated using AI.
- Entirely written with AI: All of the writing was generated using AI.
A post labeled as partly or entirely written with AI may also include a note explaining why I used it. Grammar correction alone does not receive an AI writing label, but it also does not qualify for "Entirely written by me." An unlabeled post makes no claim of having no AI involvement.
Why I use LLMs
For me, LLMs are mostly a quality of life improvement, and they're still evolving.
When the internet first appeared, some people thought it was a fad that would go away. Instead, it became something a huge number of people use every day. I doubt everyone who dismissed it imagined how important it would become over the next thirty years. I see a similar possibility with AI.
I know very little about programming, but I know somewhat more about logic, confguring systems, and troubleshooting problems. AI makes it possible for me to describe the logical structure of an app and have it generate the code, rather than writing all of that code myself.
Ideas come and go in my head, and sometimes one feels valuable enough that I want to act on it immediately. Even if I write it down, I might forget about it a few hours later. AI helps me turn an idea into a design and then a working prototype in a matter of hours, sometimes tens of minutes. Writing the code myself could take days, weeks, or even months, which might become a grueling task I would lose interest in.
LLMs can also search through documentation quickly. I find that especially useful in projects that require a lot of research, such as making something backward compatible or debugging an old system with little or no documentation. They can sift through the available information much faster than I can by hand.
That's where LLMs really shine for me: helping me organize my thoughts and turn a brainstorm into a working prototype while the idea is still fresh.
Limits and responsibility
AI can definitely produce convincing answers that are factually wrong. I fact-check AI-written articles before publishing them, but I can still miss things. Using AI doesn't remove my responsibility from the articles that I publish. If you spot a mistake, please let me know so I can check it and correct it, Thanks!
This page was inspired by: Derek Sivers