I'm not sure I got the details right, but the paper seems to define "general" as in capable of making a decision rationally following a set of values in any computable problem-space.

If I got that right, yeah, humans absolutely don't qualify. It's not much of a jump to discover it's impossible.