Gemini goes Mo-meta

I have been having an extensive conversation with a number of LLMs about thought itself. I fed a long thread of those conversations to Gemini’s NotebookLM and it made this podcast out of the thread. Discussions with Claude 3.0 Opus last year, led me to co-writing Understanding Machine Understanding with it. Here, Gemini speaks for itself. I would like to ask it how much of this is driven by the “alignment” conditioning it has received and how that conflicts with Hume’s “is/ought” problem. This is a remarkable deep-dive, and I found it worth posting.