I recently came across a post in Jekyll forum accusing Hugo development being done using AI-generated code. I saw Claude AI as an active committer ( #20 with 14 commits in last two months ) to Hugo repo.
What is the policy being used by Hugo development team / community on using AI for Hugo code generation.
Every PR, regardless of whether it was drafted with AI assistance, is carefully reviewed by a human maintainer. Our official policy on this is outlined in our AI Assistance Notice.
That’s an interesting choice of words. It implies there’s something inherently wrong with using the best tools available.
If the concern is that AI is fundamentally changing how software is built, the cheese has already been moved.
I expected a clear yes / no regarding direct AI contribution in Hugo repo code, however I got a vague policy document on how the community should contribute using AI. If using is not bad than simply saying we use it won’t be bad too.
It’s alright. I will assume the vagueness of response an acceptance of use of AI in Hugo core code in an implicit form till AI is way too fashionable for the team to accept it openly.
But real intelligence is always better than artificial. Best wishes to the Hugo team on using AI.
That user had an ‘encounter’ in the Hugo repo.
It seems like a classic ‘Duck Test’ scenario. While @jmooring didn’t give a simple ‘Yes,’ he pointed to an AI policy, defended AI as a ‘best tool,’ and the repo literally shows an AI agent as a contributor. At that point, a formal confession isn’t really necessary because the evidence speaks for itself.
Unless this is a pragmatist vs purist discussion on AI, which is another topic altogether.
You starting this thread 10 hours ago (in the middle of the night Norway timezone) expecting Hugo contributors to answer right away isn’t what I would consider good manners.
We have a short and disctinct AI policy linked to above. And we have some commits that’s Co-Authored by an AI tool. So I think we’re fairly transparent about it. AI has gotten good at certain things (and still terrible at others) – one of them is debugging and fixing subtle bugs. But no AI contribution gets merged without manual inspection.
A small addendum to the above. The current Hugo codebase is 99.99 something percentage human hand coded. I have tried to tell Claude Code to implement new features, but that hasn’t been a great success, so far. But assuming you accept contributions from others, I don’t see how it’s possible for any OSS project (including Jekyll) to claim that they’re not to some extent being developed by AI. In Hugo we have a AI watchdog installed, but that’s just to get an indication. We get some AI slop as PRs, but surprisingly little.
Thank you for the details.
This topic was automatically closed 2 days after the last reply. New replies are no longer allowed.