Matt Shumer

Is anyone trying Jev for scalable oversight? Seems like it could be useful for helping determine if a model is behaving in an aligned way.