Arno Waschk via Discussions on LilyPond development <[email protected]> writes:
> This seems wrong at least at two points: > > > 1. The automatic assumption that any support of AI makes a contributor > not read his contribution, not understand his contribution, and the > whole thing ununderstandable to a degree which can not be remedied in > the same way as hand written code is completely irrational. Both Linux kernel lists and security mailing lists have needed to change their manner of operation because they were swamped with "contributions" of people who neither understood nor vetted the AI-generated contributions they submitted. People like to "do good" by letting their AI access generate "something useful". That is quite understandable, but it does lead to unvetted AI contributions being an actual problem rather than a "completely irrational assumption". So forming a strategy to address it before opening the floodgates is not paranoia. > AI code can be bad. Manual code can be bad. I produced both > kinds. Reasonable people read their code before submitting, or, say, > try to put it in shape. Assuming people lose their reasonable minds by > using an AI summary of a local MR index is, sorry, ridiculous. The problem is not with long-standing contributors suddenly going all-in on unvetted AI. The problem is that the ability of using AI enables large classes of new contributors with different actual credentials. The best use case would be if AI were not prmarily used by newcomers for _writing_ contributions but for _vetting_ contributions. That could likely lead to an increase in the quality of contributions and less work by others until they are in an acceptable state. More neutral is using AI extensively to shape and create contributions under supervision of the creator. Likely not helpful for the workload of existing developers is to dump largely unvetted AI creations. > 2. What you wrote basically means, that code can be trusted blindly > once it was written by hand. This is but a joke. Can we try avoiding the use of hyperbole for denigrating other list members? Disagreements are enough work to sort out without artificially inflating them. -- David Kastrup
