I’ve written a bit about my interest in using local artificial intelligence for memory recall, and this week I finally made some progress on a project to start turning some of my earlier thinking into an actual part of my workflow. Memory Cache is a project that allows you to save a webpage as a PDF document, which is then used to augment context for a local instance of privateGPT.

I ran a Not-Scientific-Experiment using everyone’s favorite liar, Google Bard, to get an example of what “re-projection” for AI responses might look like in a very basic form. While the Bad Experiment above doesn’t showcase the full potential of re-projecting algorithmic responses, it hints at something more to be uncovered. What if we built a dedicated AI application that was intentionally crafted to respond with not one answer, but with many, each response filtered through prompts and datasets that reflected a specific lived perspective?

The term “communicative intent” describes the underlying message that a human is trying to get across when they are speaking with / writing to / generating art for another human to interpret. As humans, we build on our existing models of the world and our experiences to interpret one’s communicative intent when we engage with another person. One of the challenges with large language models and artificial intelligence is that there is no fundamental communicative