main thing preventing us from having a productive AI safety conversation is identity. too many AI people want to be seen as intelligent by AI safety giants. too many skeptics want to be seen as critical of EA. you have to actually think. that is the only way we'll get there.