This is a fascinating article not just about AI and the claims of AGI. It also uncovers (to me) a gap in how people are trying to evaluate “intelligence.” The rational tendencies of those developing AI think that clear benchmarks with numerical measures will say when a program has crossed a line into a “general” intelligence that is measurable. The reasoning of everyone else is that “general” intelligence is sensible and sensable. This is a fine way to live and react to the world that most of us do, but there is a dominance in the commercial space of rational thinking which rejects that as “soft” and would rather accept a poorly applied number than an accurate sensibility.
The second thing is that epistemology should be the new dance craze of the next decade. I sure as hell hope so.