Gordon Seidoh Worley comments on Is instrumental convergence a thing for virtue-driven agents?

[Gordon Seidoh Worley]

How do you get something to take virtuous action without optimizing for taking virtuous actions, and how is this different from optimizing for virtue?