Optimizing a Loss Function in a Neural Network

One thing i would really be mino cautious is relative to the cross domain knowledge transfer. You never know how thata knowledge has been transferred, in fact, from one domain, which is fighting, after all, into another domain. But definitely many risk, too many risks, before deploying a generalist agent in an uncontrolled environment or with humans.