@HarveyEsq Which part of his argument demonstrates impairment?

1.) we can make something smarter than ourselves
2.) that thing might have unexpected goals, since we can't really understand how it works
3.) there are lots of reasons why a superintelligence would kill us with and/or without intending to

To me, I feel like (1) is a lot farther away than he seems to think it is, but I don't know of any great analysis of that question to link to.