3 ms·
> At the same time, however, my experience over the past few months suggests that for system programming, LLMs almost never provide acceptable solutions if you
by 082349872349872 3y ago
> At the same time, however, my experience over the past few months suggests that for system programming, LLMs almost never provide acceptable solutions if you are already an experienced programmer.
Hmm, this suggests to me that in a better world, the systems problems would have been solved with code, and the sorts of one-off problems which current LLMs do handle well would have been solved with formulae in a (shell-like? not necessarily turing-complete?) DSL.
- chii 3y ago> LLMs almost never provide acceptable solutions if you are already an experienced programmer. or, the other stuff GPT was producing is just as bad, but that he's not experienced enough in the domain to see it, where as the stuff he is experienced with looks immediately sus or subpar.
- 082349872349872 3y agoAfter looking at https://github.com/antirez/simple-language-model/blob/main/plot.py https://github.com/antirez/simple-language-model/blob/main/p... that does seem likely.
- antirez 3y agoYes, but this is working. I literally didn't look at the code... but at the result. In other fields GPT can't deliver working code at all.