The ceiling and floor for outcomes of option D are extreme. Going to university is a good hedge by raising the floor of potential outcomes. Plus, there are some non-negotiable things like having an income above a certain level - something that uni will help with, albeit after some time and school fees. 

Something else to consider is that you're based in Taiwan. This means you likely have strong command of Mandarin and could have access to Chinese universities. China is a big player in the AI space, and in terms of international relations, AI and the tech arms race is a frontier between the USA and China. I would reconsider whether the pursuit/signalling of post-grad studies in the USA is the most comparatively advantageous path for influencing AI risk. 

I resonate with your second point, Michael. It's easy to accept skill shortages as claimed by word of mouth (we trust people we know), but more persuasive evidence is the ease of finding a job - the actual market. I also presume that there is relatively high matching efficiency for EA generalist jobs already as the 80k job board is well-established in the space, meaning that employers are finding the right people already. 

That's not to say that I don't resonate with Emily's message - I do, and am similarly navigating a shortage of EA generalist roles to apply for.