Well, for the time since then, the LLM providers have a corpus of every code suggestion they made to users and whether it resulted in positive or negative sentiment afterwards - which is arguably even more powerful. There's still some level of RLHF that's more prominent for popular languages.
As you noted, of course, this doesn't apply to architecture. But that's also why I try to make sessions as turn-efficient as possible - you need every bit of context to get it to solve its own architectural rabbit holes.
As you noted, of course, this doesn't apply to architecture. But that's also why I try to make sessions as turn-efficient as possible - you need every bit of context to get it to solve its own architectural rabbit holes.