My stance on AI writing and coding

The previous iteration of this site, before the Astro rebuild, had a notice on the footer that read:

All content (including code) on this site, attributed to me (Gonçalo MB), was created without the deliberate use of AI tools. I’m making a commitment to keep it that way.

That is still true today.

I’ve used AI tools before, but mostly as a way to search for information, both ChatGPT and Google AI Mode (Gemini). But I’m not an AI/ML developer. I don’t keep up with the latest AI news, models, tools, APIs, etc. And if you ask me about some AI thing, I probably won’t know what you are talking about.

I concede that AI is a very powerful and influential technology, but in addition to all the good that it might bring, there are also many well-documented problems, both technological and cultural.

I don’t want AI to permeate my life, and especially I don’t want it to control my ideas, creations, and memories.

Writing

I’ve never used any AI tools to write or rewrite text, blog posts, comments, emails, notes, etc. I find that idea repulsive at best. If I’m publishing some text with my name attached, I want it to be my words, not those of some machine. My writing might not be the best, but it’s still mine.

I do use normal spell checking to catch any typographical errors, but that has been the standard for decades.

If for some fringe reason I have to post any AI-generated text anywhere, it will be clearly marked as such.

Coding

I’ve always considered coding a creative thing, by this I mean that, for me, code has expression and meaning. It carries the ideas and feelings of its creator, like writing. I can go back to some code that I’ve written and see the intent and ideas behind it. That’s not to say that I consider all code meaningful, I’ve written a lot of purely mechanical code, just to “get it done”. And I know that other people just see code as “a means to an end” and that’s fine.

Also, every programmer knows that, in addition to creating new code, programming requires a lot of searching, studying, reading documentation, tutorials, etc. And sometimes we even directly reuse code and algorithms created by others. For me, up until now, this was the only way I ever programmed anything. But clearly with these “new” AI tools, that is no longer the only way.

I’ve tested AI coding tools before, and I’ve asked ChatGPT for some throwaway snippets of code. But, as of today (2025-12-12), I have never published or committed any AI-generated code to any repositories (public or private) that I have contributed to. This might be about to change.

After much reflection, I have enabled GitHub Copilot on VS Code. These are my commitments now:

Any of my commits with AI-generated code/assets (on any project) will have the author/committer email me+ai@g...mb.com (my normal git email with a +ai sub-address) and additional information in the commit message or comments if necessary. Any of my own projects that include AI-generated code/assets will have a notice (e.g., in the README.md or website).

I will consider code/assets to be “AI-generated” if they were created using AI prompts, AI rewrites/refactoring, “vibe coding”, etc. As for AI “inline suggestions”, I will consider these “AI-generated” (and mark the commit) if they don’t match what I was already going to type (in content, format, style, etc.). Basically, if I don’t feel like I personally created some piece of code, I’ll mark it.

I will also never publish any code that I didn’t read and understand completely.

With these clear commitments, I hope to be able to use these tools and continue to enjoy programming while maintaining control and authorship of my code. I’ll probably continue programming the classic way (i.e., avoiding AI prompts and “vibe coding”) but use AI as an enhancement/alternative to search and for boilerplate code while using AI “inline suggestions” as a way to be faster and more productive.

TL;DR

Writing: no AI-generated text under my name anywhere, any AI-generated text will be clearly marked.

Coding: limited AI use, any of my commits with AI-generated code will have the author/committer email me+ai@g...mb.com, and any of my own projects that include AI-generated code will have a notice.

If my stance on these issues ever changes, I’ll update this post (keeping any previous text).