Note: I was told by a friend to cross-post this from Substack. Let me know if I'm violating any community norms!

“Here is one common failure mode for EAs. You look at 80k’s list of the most important cause areas, pick the first one as the thing you want to work on, then end up doing that.”

Oh. This sounds like a familiar story.

* * *

It’s summer before college, and I’m staring blankly at the worksheet in front of me. I’m a Non-Trivial Finalist — not a Fellow yet, though I intensely want to be — and this worksheet is my last barrier to entry. The topic of last week, Week 1, was “Clarify your mission.” The topic for this week is “Choose a pressing problem to address.” Easy peasy.

All I have to do is pick a cause area and a project I want to work on in that area. I’m more interested in brainstorming a project idea, so I rush through the cause areas. AI safety is the first thing listed on the 80k website, and I kind of like math (even if I know nothing about AI), so it seems like maybe this is something I could be good at. I don't struggle to find reasons to rank AI safety as my top priority, especially because the other problems feel so foreign (biosecurity?? I napped through 9th grade bio), and the arguments in favor of it are pretty compelling. But it’s easy to make compelling arguments.

* * *

I imprinted on AI safety like a duckling imprints on the first moving object it sees. But what if my mother duck is a goose or a chainsaw or something? Indeed, I realized today that I’ve never thought very hard about cause prioritization. This might be a big mistake.

When I say cause prioritization (which I will henceforth refer to as cause prio because few word do trick), I mean, “reasoning about what problems are most impactful for you to work on.” One strong argument in favor of spending a lot of time doing cause prio is as follows. You will spend around 80,000 hours of your life working as part of your career. It’s worth spending even 1% of this time thinking about what problems are most valuable for you to work on if you care about maximizing your impact. I have definitely not spent 800 or even 100 hours thinking about this.

In the same conversation that made me realize I never really thought about cause prio, I was told that I should know the arguments for AI safety front-and-back. And not just for AI safety, but also for other cause areas that seem worth thinking about, such as animal welfare, protecting democracy, and whatever else. I was a little unconvinced about this at first. In particular, my gut reaction was, “I’m already quite bought into AI safety, and I’m scared we live in a world with short timelines. Is the opportunity cost of thinking about cause prio part-time over the next few months too much?” A quick BOTEC seems like it could help with answering this question, but the answer seems like no, thinking about cause prio is still good on average. In particular, it seems like 3 extra months of AIS work from me make a small difference in the short timeline, but 3 extra months of cause prio make a potentially big difference in the longer timeline.

Upon closer investigation, it seems like my initial reluctance to take cause prio seriously was because I didn’t want to feel lost and confused. I wanted to hold onto the feeling of purpose that was recently instilled in me by believing more in the importance of AIS. I also have obvious motivation to believe in the importance of AIS, since it’s looking like I’m doing something in AIS over the summer and will likely be leading PAIA next fall, and I don’t want to regret the effort I put into those endeavors. But I comfort myself with the thought that, if not the most important cause, AIS is probably still pretty important, and my contributions will be worthwhile if I try to make them worthwhile.