This is a series of post in which I laid out the foundations of what would become the research agenda of the Phenomenological AI Safety Research Institute (PAISRI). Originally posted on the Map and Territory blog on Medium and linked on LessWrong, they have now been copied over to LessWrong to make it easier to engage with them and host them on a site with a stronger commitment to hosting AI safety related content long term.

I no longer necessarily endorse everything in these posts as I have learned more, but I continue to endorse their spirit and consider them worth reading.