News for hackers. Stuff that matters.
Re:Hallucinations in code are the least dangerous form of LL... by tokioyoyo on Sunday March 02, 2025 @07:51PM (#43237101)
The big argument against it is, at some point, there’s a chance, that you won’t really need to understand what the code does. LLMs writes code, LLMs write tests, you find bugs, LLM fixes code, LLM adds test cases for the found bug. Rinse and repeat.
Re:Hallucinations in code are the least dangerous form of LL... by SamPatt on Sunday March 02, 2025 @08:28PM (#43237342)
For fairly simple projects built from scratch, we're already there.

Claude Code has been doing all of this for me on my latest project. It's remarkable.

It seems inevitable it'll get there for larger and more complex code bases, but who knows how far away that is.

Re:Hallucinations in code are the least dangerous form of LL... by saagarjha on Monday March 03, 2025 @05:06AM (#43240215)
What do you do when the LLM doesn't fix the code?
Re:Hallucinations in code are the least dangerous form of LL... by amarcheschi on Monday March 03, 2025 @07:46AM (#43241221)
You tell it there's an error, and to fix the code (/s)