I followed the link to the tweet thread where the graph was found and it’s not much better

https://threadreaderapp.com/thread/1666482929772666880.html

A concrete “easy scenario”: LLMs are just straightforwardly generative models over possible writers, and RLHF just selects within that space. We can then select for brilliant, knowledgeable, kind, thoughtful experts on any topic.

Like, wtf does this mean.

Another recent HN title

“LLMs are too easy to automatically red team into toxicity”

Again, wtf does this mean.

I’m increasingly wondering whether I am officially too fucking old for this shit, and my brain has lost the plasticity required to make sense of all this, or I’m just spending too much time on the wrong Internet.