I am thinking of regularly writing a Lesswrong sequence on Causality applied to Machine Learning, as a lesser alternative to doing an AI Safety postdoc on the topic I suggested here.

The purpose of both would be to learn sufficiently about Causality applied to ML such that I can later on contribute with original research. For context, I am finishing a Ph.D. in quantum algorithms so I know how to do research, the issue is learning about a new research area rather.

However, the postdoc, which is still dependent on the OpenPhil decision next week, seems a bit worse career option than a job as a quantum research scientist in a startup in some respects. The main reason for that is that my girlfriend would prefer the relative stability of a well-paying non-academic job in a field I already know, and the ability to work from home. Some have also argued that having one expert in quantum computing in the community might be useful, in the unlikely case it becomes useful, and it would also help me partially skill up as a ML software engineer.

In any case, I would still like to learn about Causality applied to ML and AI Safety because I think there is insufficient research going in this direction in the community; so I was thinking that as a way to learn about it, I might regularly summarize papers or something similar. Lesswrong also supports $LATEX$ which makes it convenient as compared to other platforms.

I was thinking of writing a sequence about this topic to learn. How should I go about it, if I intend to work on this during weekends, say 1 day/week?

Thanks!