Rohin Shah comments on Some AI research areas and their relevance to existential safety