@cadey Ugh, I just realized I have to watch the video before replying. I'll see if I have some time later.
@cadey (And to be clear, the "ugh" was directed at me, for failing to realize there was a video!)
@cadey Yeah, the text was pretty accurate. I don't know, we haven't solved CRUDs in... how many years of writing CRUD apps? I have not seen LLMs automating CRUD.
My metaphor is that LLMs are good at "creating stuff that *looks* like X". So I'm concerned about, say, creating illustrations, which is really "creating a picture that looks like X".
@cadey But creating source code that looks like the solution to problem X (which LLMs are pretty good at doing!) is not creating source that solves problem X.
I do think they're capable at interpolating SO answers into a solution, but normally I have to do a lot of stuff which I cannot do by interpolating from SO answers.