We are still trying to solve the problem that we can't keep the plasma hot long enough to create fusion energy, so working on exotic conversion schemes is one step too far.
Consider also how complex these reactors already are, it makes sense to use the simplest method that we know works well.
Makes sense, but from a layman perspective it seems like introducing additional complexity and lots of inefficient, high-loss transmission steps.
We start with detached electrons moving at high speeds (plasma). We want detached electrons moving at moderate speeds (electrical current). And yet, the intermediate steps involve everything from heat, steam, large-scale mechanical forces and magnetic induction, just to get back to the electrons?
It feels more like the "pull in a 500MB framework instead of writing the function yourself" kind of simplicity.
Consider also how complex these reactors already are, it makes sense to use the simplest method that we know works well.