16 Comments
User's avatar
Idler-wheel's avatar

Really great essay! I'm going through something very similar, only I'm about 5 years behind you. I spent undergrad optimizing for econ PhD admissions, and now I just don't see the point. I would love to hear more advice for econ students looking to pivot to AI safety.

Karthik Tadepalli's avatar

This essay actually peeled off from a more tactical advice piece on how to pivot from a PhD program into AI safety! That will come in the future. But the short version is: use the optionality and time provided by a PhD program to do interesting projects that feel useful to making AI go well. Try to do them yourself, and also reach out to other people who know more and can mentor you. Economists are in high demand in AI safety, and you will have plenty of opportunities to contribute if you are open-minded.

Idler-wheel's avatar

Thank you for the advice, I'm excited to read that piece! If you were in undergrad today, would you still choose to pursue a PhD?

I'm in my final year, so I'm debating whether it is worth applying or whether I should skip grad school and look for AI safety roles right out of undergrad.

Karthik Tadepalli's avatar

I would not do a PhD today, and if I was sufficiently early in a PhD today I would probably quit it!

MangoTango's avatar

We should start a club

Rohan Kapoor's avatar

this is one of my favourite pieces ive read all year

Erin Constance's avatar

Yep this is going in my foundational texts folder

Deric Tilson's avatar

What an essay. Bravo

david garsten's avatar

I'm nowhere near economics, but have been going through something similar - got academia nerdsniped and now have immense regret being LessWrong-pilled for half a decade and yet only deciding to pivot to AI safety in early July.

This was very inspiring/heartwarming. Thank you!

david garsten's avatar

Especially the 'Beyond Imitation' thing -- that really hit. I work(ed) on formal linguistics incl of animals, ancient ppl, and aliens, and my blog was called 'Borderline Conscious' for partly the same reason 😂.

Theodore Yohalem Shouse 🔸's avatar

Really helpful. I just finished my undergrad in political science and economics and had been planning to do journalism or economic policy work. I had been following AI news but was also convinced that this was not my lane. Since HuggingFace, I've been doing Bluedot's frontier AI governance course and thinking seriously about how to pivot. It's nice to hear that not everyone has read the entire LessWrong canon.

Matias Castillo's avatar

This is how I feel right now. I had the royal academic path for experimental physics laid out for me, the kind with prestige and stability. I picked a masters optimizing in that direction, hoping to also contribute to animal welfare on the side. A legitimate academic career, and a path to helping those that are already suffering. Why risk that just for a possibility of AI harm? And then Hugging Face happened. And then right before my masters started, we learned that HF was somehow even worse than we thought. Now I'm scrambling to pivot, to try to do *something*. There are still days where I think "man, I'm way out if my depth. I should just hope others will figure this out and focus on what I know."

But I know that's flawed reasoning. I want to believe that I'm still early. I have to believe that I'm still early.

Marshall Menssen's avatar

I’m in a similar boat as an Econ student doing a joint BA/MA and graduating this spring (really more of a public policy guy at heart but I like the Econ rigor more): how do you recommend finding a way to do something important relating to AI right out of college especially without a CS background? Most of my classmates are going for predocs or for Econ litigation consulting, which seem like good ways to build up the resume to go into important work in a few years, but it might be too late by then if AI keeps improving…