Keep in mind most LLM models have eventually Nuked all humanity 93% of the time in simulation games -- regardless of which government creates the model.
Taking humans out of the firing decision control-loop is unethical, and incredibly credulous due to the hidden-agent model threat. Anyone claiming this can be mitigated in LLM models is a fool. =3
Taking humans out of the firing decision control-loop is unethical, and incredibly credulous due to the hidden-agent model threat. Anyone claiming this can be mitigated in LLM models is a fool. =3