[FuriouslyAdrift]

An LLM is just a really good parser connected to a lossy compressed corpus of data.

They need to be open ended and self training to be truly useful.

Reasoning is way far away...