LLM generated code has definitely improved in quality over the past year, but I still think it is very easy to spot when code is not written by a human.
And I'm not talking about the super obvious signs that everyone already knows like the overuse of emdashes and such.
What I'm talking about is that LLM generated code is just really fucking boring...
Surrounding literally everything with null guards, strict type checks everywhere, throwing exceptions for every little minor thing, etc.
I personally love weak comparison. It is fun, feels hacky and it doesn't produce insecure code if you actually know what you're doing.
The LLM has to burn a million tokens to consider every single possible "but what if ..." situation when doing something like this which it will try to avoid.
A good programmer actually familiar with their own codebase will just look at the code and say "no, it wont" and will delete the overly verbose null guard or whatever.
How a Bad AI Camera Hit Put the Wrong Man in a San Diego Jail Cell
https://reclaimthenet.org/san-diego-flock-misidentification?utm_source=fediverse
@ube@tsundere.love @hj@shigusegubu.club @Phobos@eientei.org @nicole@kiwifarms.cc Sometimes kiwifarms is iseful though. Like when they host material mainstream media doesn't want you to see. Or when they dig up dirt on people who actually deserve it like Drew DeVault.
So virtually every video game released after the SNES is potentially infringing copyright in some way due to the legal gray area of soundfonts.
https://inv.nadeko.net/watch?v=wR2aIzh2_fc
To add insult to injury, check out this comment of the same video:
@sendpaws@mitra.pawslut.party Forced to use a real computer instead of phonepost? I don't see the issue
Turns out eating the bugs is a genetic inheritance that the enlightenment cultures don't have.
That said, it does seem they got an extra dose of the LIVE IN THE POD gene.
https://bird.makeup/users/is_owenlewis/statuses/2063405676324626511