Perhaps it’s ironic to use an LLM to analyze Benedict’s presentation on LLMs. But if there’s a way to dig through a comprehensive aggregation such as Evans provides, it seems LLMs are one good way to do it. Thus the analysis below.
Might be interesting to see what a different LLM such as Gemini 3 might make of Evan’s presentation too.
At some point, information requires forms of opinions, and subjective takes. It’s not all just naked “facts”.
Information needs interpretation, and that often involves one opinion vs some other opinion. That’s a challenge we will likely encounter soon if we follow an LLM platform agnostic approach, and use a variety of models for the same task, as LLMs do seem to somehow have opinions of their own.