| Filename | Latest commit message | Latest commit date |
|---|---|---|
Seth Falco
21d1212b12
All checks were successful
ci/woodpecker/pr/ci Pipeline was successful
I shared the original AI policy / document with some friends for feedback and receives two criticisms: 1. The prominence of the word "never" (given the phrase context in which it was used) was undeserved. I reduced the scope of what the "never" applies to. 2. My use of the word "voluntarily" was wrong. I treated the word to mean I wouldn't instigate, encourage, promote, or table the idea to use AI for these purposes. However, it's still a voluntary use of AI if a client offers money and I choose to accept it. |
||
| .woodpecker | ||
| .editorconfig | ||
| cspell.yml | ||
| LICENSE | ||
| README.md | ||
Generative AI Policy
TL;DR: My Open Source contributions, personal projects, and client work are all human-crafted by default.
For personal endeavors, I will never:
- Give you AI-generated code, documentation, illustrations, music, etc.*
- Publish or sell AI-generated artifacts.
- Use generative AI to pursue creative endeavors that replace human creativity.
- Communicate with you using AI—this includes "organizing thoughts".*
- Vibe code, or anything adjacent to it.*
I only ask that you show me the same courtesy!
* For professional work, I don't necessarily refuse when clients solicit me to use AI in this way for their projects… mortgage and all, y'know?Why?
I value quality, craftsmanship, and human ingenuity. I want an intimate understanding of the work I publish, and to acknowledge the human contributor(s) who I place my trust in when incorporating their ideas into my work.
The process of creating products should be thoughtful and intentional. And the products themselves should be accessible, performant, and perhaps have a touch of whimsy. Generating code and assets en masse misses the mark in my opinion—my rationale isn't very different from why I don't use WYSIWYG editors to build websites either.
But you have AI projects?
Indeed! Ironically, not a single one of them is AI-generated.
This organization has projects of mine which tie in with generative AI, and may even support use cases that I don't agree with myself. I'm still going to take pride in my work and share it under a free-culture license.
Do you voluntarily use generative AI?
I self-host Ollama and Hermes Agent, and experiment with models that are compliant with the Open Source AI Definition like:
- Apertus by ETH AI Center & EPFL AI Center 🇨🇭
- Olmo by The Allen Institute for Artificial Intelligence 🇺🇸
I also work for clients that may provide subscriptions to the typical frontier models. When I have access to these models anyway, I do set them up for personal use as well.
So yes! Some examples of how I use generative AI:
- Debug issues in code I've written.
- Check code or articles I've written for obvious mistakes.
- Search for answers when the documentation is too sparse.
- Identify open bug reports or feature requests that are already resolved.
- Investigate obscure system errors or logs.
Won't somebody please think of the ethics?
I'm not dismissing ethical concerns, the toxic culture throughout the industry, and how fucking annoying it is to see ✨ AI ✨ everywhere.
I just think it's healthier to stay engaged and support organizations that will steer AI in a better direction—not to disengage and discover that only the unethical conglomerates survived the bubble.
*shudders in decades of schools and workplaces only using Outlook or G Suite*
Want to talk about it?
If you have questions or concerns about this, get in touch! 💪(•ᴗ•💪)