After talking to jsalvatier offline for a bit about the issue of wanting to do AI research, but being afraid to fail, I've came to realization that I haven't fully analyzed why I am afraid of it. I'll need to take more time to break it down, but my first guess is simply lost time. But it's very likely that if I try to do the research, I'll realize really fast if I am at all capable of doing it, so the time wasted would be a year or two, which is not too bad IMO.

You said you are interested in existential-risk reduction. I don't know if you mean specifically AI research by it, but if you do, we should try to solve this problem together. I am sure there are other people in similar situation (and there will be even more in the future), so it makes sense to create some kind of standard way of answer the question "am I cut out for this".