This is a special post for quick takes by micseydel. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
I'm working on a project that may be automatically aligned in some sense. Thinking in terms of the Thousand Brains hypothesis, I have a network of virtual, specialized cortical columns that help track different reference frames for me with a wiki shared between myself and these virtual columns to act as a cognitive glue. The virtual columns are like "atomic" agents that center linked Markdown notes, styled off of "atomic" or "networked" notes PKM styles. These atomic agents are not prompts right now, they're mostly Scala code, an explicit encoding of my agenda. If any AI system seems like it can help my agenda, I can incorporate it by wrapping it with an actor. Right now that includes Whisper for transcription and Rasa for entity extraction.
I need to incorporate Bayes' law explicitly, and I only recently started adding voting a la Monty (the main Thousand Brains project), but I thought it would be worth making some post here just in case it catches anyone's attention.