I'm pretty sure someone sent another LLM-generated patch to librsvg.
-
I'm pretty sure someone sent another LLM-generated patch to librsvg. It's for a feature in rsvg-convert that I don't want, so I explained my rationale as kindly as I could.
But the code is very subtly wrong. I did not comment on the code at all.
-
I'm pretty sure someone sent another LLM-generated patch to librsvg. It's for a feature in rsvg-convert that I don't want, so I explained my rationale as kindly as I could.
But the code is very subtly wrong. I did not comment on the code at all.
You know that painstaking, standard-looking format for a merge request that LLMs like to generate?
Summary
Motivation
Implementation
Files changed
Usage
Patch
Am I being unkind in ascribing LLM usage due to *that* format? If LLMs didn't exist and a person had written that, I would say, well, how organized and kind of them. But I know no one writes MRs like that, and I feel conflicted about that being my main source of suspicion.
-
You know that painstaking, standard-looking format for a merge request that LLMs like to generate?
Summary
Motivation
Implementation
Files changed
Usage
Patch
Am I being unkind in ascribing LLM usage due to *that* format? If LLMs didn't exist and a person had written that, I would say, well, how organized and kind of them. But I know no one writes MRs like that, and I feel conflicted about that being my main source of suspicion.
@federicomena It's interesting to me that an AI would produce a MR like that, given that, like you say, no one writes MRs like that. So where the heck did they learn to use that format?
And I fundamentally don't understand these LLMs. I just feel like they're vaguely complicated, and I have no idea what they're doing or how they're doing it.
-
R relay@relay.publicsquare.global shared this topic
