As I have previously mentioned, I'm on the PC of POPL this year. There were an unprecedented number of submissions (>600), which meant that even with reserve reviewers, each PC member ended up with 14 papers to review. Seeing as POPL papers are typically dense with a fair bit of maths, this is an extreme amount of work.
My workload was exacerbated by the fact that the Toronto paper matching system used by POPL seemed to produce ranks worse than random, so even before being assigned reviews I had to examine (at least the titles of) every submission to the conference to figure out what to bid on.
I don't think we want our PL and FM conferences to end up like AI conferences, with enormous PCs and an even higher volume of papers. I don't want our research to turn into a paper mill where we throw papers over the wall that even the reviewers barely read, let alone anyone else (at best, a disgusting AI summary might be all that is left of our papers for human consumption). But if things keep progressing this way, that's exactly where we'll end up — the volume of papers even in this iteration of POPL is such that I'm having to take shortcuts just to get the reviews done in time.
Normally, my policy is to write a review, wait a day (while not reading anything else), and then submit it, because often I will keep thinking about the paper while I go about my day. This PC membership has made that impossible.