Ellie via Postfix-users:
> Dear mailing list,
>
> My apologies if this has been answered before or I just didn't look at
> the website enough, but does Postfix have any sort of LLM/AI policy for
> contributors?
[snip]
> (I'm less curious about whether anyone cares regarding passive use like
> LLM code analysis, and mainly wondering if there's a policy regarding
> actually submitting LLM-produced code snippets directly into the code base.)
The only policy is that nothing goes into the code base without
human approval.
AI, like a human contributor, may come up with an inappropriate
solution because they (AI or human) aren't aware of some problem
nuance. AI saves time implementing candidate solutions, but that
is only one step in the development process.
In Postfix, LLMs have recently assisted with:
- Code review, in addition to human review.
- Code refactoring (mysql/pgsql/sqlite clients, no change in
behavior), followed by human review.
- Adding an incremental feature (the first iteration of TLS trace
logging was implemented by AI; that result was then simplified
without using AI during human revision and review).
Wietse
_______________________________________________
Postfix-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
