Debian dismisses AI-contributions policy

Posted May 12, 2024 14:41 UTC (Sun) by farnz (subscriber, #17727)
In reply to: Debian dismisses AI-contributions policy by james
Parent article: Debian dismisses AI-contributions policy

That leads into the other problem with a "no AI contributions" policy. Imagine I take my code, feed it to an LLM, and ask the LLM to write documentation for my code. I then take the LLM's output, and rewrite it myself to be accurate to my code, checking for mistakes; I used it as a prompt since I struggle to get going on documenting things, but I find it easy to fix bad documentation. The result is something that reflects considerable effort on my part, and that has very little trace of the original training data that was fed into the LLM; the LLM's role was to show me what good documentation would look like for my project, so I had a goal to aim for, even though I could not reuse the LLM's output.

Is this a contribution that we should reject? If so, why?

Debian dismisses AI-contributions policy

Posted May 14, 2024 1:06 UTC (Tue) by mirabilos (subscriber, #84359) [Link]

Yes, as it’s still a derivative of something that isn’t legal to use.

And for all the other reasons, such as model training being done in unethical ways that exploit people in the global south, and LLM use requiring too much power (and even other natural resouces like clean water, for reasons), so you should not be using them a̲t̲ ̲a̲l̲l̲, period.