[Question] Career Advice?

I’m an as­piring ra­tio­nal­ist /​ effec­tive al­tru­ist /​ math­e­mat­i­cian, on track to finish a mas­ter in com­puter sci­ence in a cou­ple of months. I’m try­ing to figure out what to do next; that’s the de­ci­sion.

Var­i­ous con­sid­er­a­tions, pri­mar­ily read­ing ar­ti­cles on 80000 hours have led me to de­cide that I want to do a PhD* in Ma­chine Learn­ing, with the goal of work­ing on tech­ni­cal AI safety, or earn­ing-to-give as a backup plan. With that de­ci­sion already made, I think I don’t want to ap­ply for 1-on-1 coun­sel­ing (ob­vi­ously I would take the spot off of some­one else), but I’m still left with a cou­ple of ques­tions that I don’t re­ally know how to ap­proach.

(1) * This is a rel­a­tively minor point, but – I live in Ger­many, and the more com­mon de­gree here is some­thing called a doc­torate. Sup­pos­edly it is equiv­a­lent to a PhD but the pro­cess differs a bit in that the doc­torate is even less struc­tured and even more fo­cused on in­di­vi­d­ual re­search. I’ve read that many uni­ver­si­ties do offer the real PhD though. Ig­nor­ing the differ­ence in the pro­cess, I’m won­der­ing whether I should or shouldn’t worry about which de­gree will be more valuable, and maybe try to find a PhD pro­gram rather than the reg­u­lar doc­torate.

(2) Ac­cord­ing to this site, my uni­ver­sity ranks #126–150 in com­puter sci­ence and #251-#300 wor­ld­wide. This is not su­per high – my ques­tion is, how much does this mat­ter? Has any­one cred­ibly tried to quan­tify this? Also, are these rank­ings even re­li­able, or should I use an­other site?

The two op­tions I’m pri­mar­ily con­sid­er­ing is stay­ing where I am or mov­ing to Ber­lin. Ber­lin’s high­est-ranked uni­ver­sity in com­puter sci­ence is at #55.

(3) Re­lated to (2), I don’t re­ally know how difficult it will be to ac­tu­ally get a PhD /​ doc­torate po­si­tion. 80000 hours has some re­sources on this, but they gen­er­ally as­sume that the goal is to get into one of the top 5 uni­ver­si­ties wor­ld­wide, and in fact they men­tion at one point that any­thing be­low that might be a lot eas­ier. Ob­vi­ously #55 is not any­where near the top 5, so ba­si­cally I’m won­der­ing whether just finish­ing a mas­ters with good grades is re­al­is­ti­cally enough, or if one re­ally does need to have pub­lished stuff already. The uni­ver­sity web­site doesn’t an­swer this; they just tell you the for­mal re­quire­ments. The an­swer here might just be “it de­pends”, but I don’t ac­tu­ally know that.

(4) I don’t like the ma­chine learn­ing in­sti­tute at my uni­ver­sity; I have is­sues with their lec­tures and par­tic­u­larly the lack of rigor. This is one of the rea­sons I’m con­sid­er­ing mov­ing. On the other hand, an­other pro­fes­sor (whom I do like) has more or less offered me a doc­torate po­si­tion that would be about some­thing like “pri­vacy-pre­serv­ing ma­chine learn­ing”, i.e. a hy­brid be­tween ML and in­for­ma­tion se­cu­rity.

I’m prob­a­bly not go­ing to take this be­cause I imag­ine it’ll be a lot less valuable than a pure ML PhD (?), but I’m not sure yet. It would be awfully con­ve­nient.

This mostly sum­ma­rizes where I am right now – any in­put is highly ap­pre­ci­ated.

Meta: I was go­ing to post this in an open thread rather than make it its own thing, but there don’t seem to be open threads any­more. I think this definitely raises the bar for ask­ing things – if one has to start a new topic, they might worry about whether it’s ap­pro­pri­ate and such, and end up not do­ing any­thing in­stead.