Usually code contributions by various LLMs are easily identifiable because the agent is the author for the git commit. Mozilla on the other hand seem to be explicitly encouraging unattributed LLM code in Firefox. Also note jakearchibald, Mozillas AI spin doctor whenever devs question their intentions, lying about the reasons for this change. I think their true intentions are to muddy the waters to hide the amount of slop contributions in Firefox.

you are viewing a single comment's thread
view the rest of the comments
[–] 133 points 7 months ago (9 children)

This change is about preventing AI from trying to own the change. A human must own the change.

AI cannot own a Firefox contribution. AI cannot commit code to Firefox. Only a human may do that.

If a human uses AI (or autocomplete / a formatter / a transpiler / whatever else) to help them author code, that doesn't devolve them of responsibility. The human must take ownership and responsibility for the output.

For example, if we later run git-blame on a section of the code, we want to see the human that took responsibility for the code, not some AI.

Firefox's policy on AI code: https://firefox-source-docs.mozilla.org/contributing/ai-coding.html

  • source
  • hideshow 9 child comments
  • [–] 25 points 7 months ago

    Perfect way honestly. Same philosophy we follow. I don't care how you made the PR, you're responsible for it and you need to be able to defend and explain it. If we get a "I don't know, chatGPT told me this" reject.

  • source
  • parent
  • [–] 18 points 7 months ago (1 child)

    How many human eyes of experienced programmers will have looked at the code?

    A whoopsie in the web browser is significantly worse than in an... I don't know... than in a vibe-coded so-called "AI" framework/social network. /s

  • source
  • parent
  • hideshow 1 child comment