Ajeya from OpenPhil wrote a very understandable and quite compelling summary of Paul's views on AI-Alignment.

(@Paul & @Ajeya: I would love to crosspost the whole thing here if you are open to that)