Apologies if this is not the best place to ask this, but I’ve done my homework and this seems the best bet.
Question: How important is having internships during my PhD (in machine learning)?
My supervisor runs a tight ship, and (near-explicitly) forbids lab students from doing internships . The expectation is that the entire duration of the PhD will be dedicated to research in the lab. The logic goes: The supervisor found your funding, so you (quite literally) owe it to work only for him/her  . I find this culture rather toxic, and wobb (click through the up coming page) so does everyone in the lab, with the exception of the one or two kids who are hell-bent on academic careers (so in their case it doesn’t matter).
I understand that doing internships is integral to experience, networking, and in general figuring out what you really want to work on post-PhD. So my question is more: If I don’t do any at all, how bad is it in terms of career opportunities after graduating? To make things worse, my supervisor also declines to collaborate with anyone from industry labs (even though groups from the likes of DeepMind have actively contacted us for collaboration). To me, this is just plain silly. So basically it’s a double-whammy (i.e. I’m not even networking in the most minimal sense).
For what it’s worth: This is a “good” lab on paper: We all manage to publish in NeurIPS, ICML, ICLR, etc. including myself, so that’s what we have going for us. (But that’s also just due to a near-abusive supervisor who whips us into doing nothing but writing papers all year round). I’m just incredibly distressed about the fact that, even if I were fortunate enough to find an internship somewhere, I wouldn’t be allowed to do it. After all, there are plenty of other candidates with 2-3 internships at the big labs, and just as many good publications, and I just worry about the uphill battle in terms of experience, network, etc. when it’s our turn to graduate.
Thank you SO much for any and all advice…! I’m worrying myself to pulp, and I know I’m not alone in this lab in doing so.
Previous graduates: Most of them ended up being professors. The ones that didn’t ended up at decent places, and even a few at Google. But back then they were allowed 1 internship, whereas now the—again, unspoken but pretty clear—policy is no internships.
Changing supervisors: This is pretty much out of the question. I’m 2 years away from graduating; if I switch now it would involve months of paperwork and unclear approval about “transferring” prior progress. Not sure if this bureaucracy is specific to my university, but someone from the lab actually did it before, and had to restart the entire PhD from the “non-probationary” point onwards.
 Effectively, this means either causing problems with your funding or candidacy, or threatening to postpone your graduation until kingdom come, not to mention the continuous guilt-tripping.  I’m in a program where you cannot “freeze” your funding for 4 months, say, while you do an internship). I’ve tried my best to explore this option, but even so, point  applies.