Yes, some tasks, even complex tasks will become more automated, and machine driven, but that will only open up more opportunities for us as humans to take on more challenging issues. Each time a great advancement comes we think it's going to kill human productivity, but really it just amplifies it.
Where this ends is general intelligence though, where all more challenging tasks can simply be done by the model.
The scenario I fear is a "selectively general" model that can successfully destroy the field I'm in but keep others alive for much longer, but not long enough for me to pivot into them before actually general intelligence.
It melds Firewall and DNS to block.