At first, I wanted to identify crucial factors that lead to alignment and then build infrastructure for that. I would start by identifying important problems, and then figure out if infrastructure would be a good solution for that.
However, it looks like there are many factors that contribute to reducing AI risk, like how policy, technical, hardware, and norms interact. So I thought I could learn how systems work and how the AI ecosystem interacts with each other.
My routine would be taking the BART to high concentration places in SF or berkeley, sitting down writing my blog, and talking to people that are around me. I could learn how systems work with books, courses, and YouTube videos. I could read papers and do independent thinking exercises to learn more about AI.
Someone told me that I would learn better by collaborating with other people and getting a job so I get structure and feedback loops. So now I think I'm going to write commentary on papers with friends and apply for jobs.