Skip to main content


Servo is considering:

- allowing some AI tools for non-code contributions

- allowing maintainers to use GitHub Copilot for some code contributions over the next 12 months

These changes are planned to take effect in June 2025, but we want your feedback. More details:

github.com/servo/servo/discuss…

This entry was edited (1 week ago)

reshared this

in reply to Servo

Cringe + L + Ratio

I don't know how not to be confrontational here why do you want your users to eat slop you're not mozilla

in reply to Servo

Relationally and technically IMO this is a bad move.

It reduces the likelihood of me building with or donating to Servo for various reasons.

Code is one of the least justifiable uses of LLMs, and then we have a plethora of ethical and potential legal problems needing to be worth any perceived benefits.

Why risk such a great project for questionable benefits?

in reply to happyborg

As much as I'm very much not a fan of using LLMs for code (or anything really), I don't have a solid answer to this question.
On the one hand, I'm pretty sure that it won't work, on the other I think a project like Servo is perfect for such an experiment, it might become a good case in point of why exactly this doesn't work — so others would stop having such ideas.
I feel conflicted about this 🤔
in reply to Servo

Very very bad idea and makes the whole project non FOSS. Rather it's an amalgamation of undocumented, undocumentable, plagiarized code of unknown provenance, attribution, and license. And likely buggy and full of vulns.
in reply to Servo

boom. You're dead.

Edit: Enshittified even before version 1. That was impressively fast.

This entry was edited (1 week ago)

reshared this

in reply to Servo

allowing AI code will cause licensing chaos. Code "created" by an AI can not be copyrighted, and thus throws your entire MPL licensed project into legal jeopardy.
in reply to Servo

don't do this, please.

The research data are pretty clear that LLM usage can improve the speed of authorship but usually at a medium to high cost in quality, which should be enough to disqualify its use; but even more importantly, adopting it will telegraph a project posture that will alienate exactly the kind of developer and user community a young project should be aiming to attract.

I hope you succeed wildly, thus I hope you do not adopt a policy of accepting LLM contributions.

in reply to Servo

The way LLM makers have been literally DDOSing our public infrastructure I think this is morally wrong.

As for using it for documentation, I think the comments on accuracy are on point. If I want to read the project's documentation I want to read accurate authoritative information written by humans that know which parts are important. If I really want an LLM to summarise the codebase for me (I don't) I can do that on my own time, no need to add slop to the project's documentation.

reshared this

in reply to Servo

i really appreciated seeing the dissent made quite clear in the steering committee minutes floss.social/@servo/1142632664…. it's not clear to me who contributes to the voting populace that led to overturning the prior policy which restricted these, and i also thought it was strange that the dissent was not resolved before passing by vote.
in reply to Servo

I know a lot of the tech community is excited for Servo after seeing how other browser engines have disappointingly embraced AI.

I don't think this is a good idea. It's both a moral and technical dilemma that isn't worth the headache.