Post #2702965

2026-04-03 01:17 UTC

@protowlf@mastodon.gamedev.place Do you have a good source (ironically) that sets out the reasoning behind why we should understand what's going on inside one of these models while code is coming out as "plagiarism"? They absolutely will spit out whole chunks of code that one can point to and say "that was copied from such-and-such without following the license". But they also don't *always* do that. Are they *always* plagiarizing, or only sometimes?

Replies (2)

@Forbearance@mastodon.xyz 2026-04-03 01:19

@protowlf@mastodon.gamedev.place I looked and found https://lawreview.uchicago.edu/online-archive/plagiarism-copyright-and-ai but that is set in the academic and legal sphere, where a whole idea or concept being spit out without the citation to its originator constitutes "plagiarism". I assume I'm not guilty of plagiarism in the computing field every time I apply a facade or visitor pattern without citing the Gang of Four.

Open ##2702966
@gray17@mastodon.social 2026-04-03 05:42

@Forbearance@mastodon.xyz @protowlf@mastodon.gamedev.place someone studied how well coding agents solve simple programming problems in esoteric languages like whitespace and brainfuck. the agents all fail at things like, "compute n factorial for n < 10". this seems to me a pretty strong indication that the coding agents don't have any significant "reasoning" ability, they mainly work by regurgitating semantically-similar-shaped things from their training data. https://arxiv.org/abs/2603.09678

Open ##2702979