Anecdotally, I know some scientists who prefer Sci-Hub even when their institution has a subscription to the journal. The login procedures (and possibly VPN setups when working from home) are too much work when Sci-Hub is immediate.
I'm sure that publishers will try and fight back, just like with music copying.
I would always prefer to get the PDF in the legit way unless sometimes I encountered journals that my institution do not subscribe to.
Only reason for that: to reduce the burden to scihub servers.
I'm sure that publishers will try and fight back, just like with music copying.