Hacker Timesnew | past | comments | ask | show | jobs | submitlogin

Forgive my ignorance, but if the corpus of coding data was always 90% bad, isn't that the same data being used for training LLMs? How are they magically any better than that average?


Programmer: "What is this slop that I found in your code?"

AI: "I LEARNED IT FROM YOU, DAD!"


They aren't. The guy you're replying to is just hyping them up based on nothing.


Because LLMs are not stochastic parrots.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: