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

I recommend you check out Andrej Karpathy’s 2 YouTube videos on how LLMs work (they are easy to find, but be forewarned they are long!). Once one digs in deeper it becomes clear why a model today might fail at the task you described.

Generally speaking, one of the behaviors I see in my day to day work leading engineers is that they often attempt to apply agentic coding tools to problema that don’t really benefit from them.



> I recommend you check out Andrej Karpathy’s 2 YouTube videos on how LLMs work

I'll reccommend Claude "AI" do that, so it knows to tell the user when it /doesn't/ work.




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

Search: