[chbint]

Their argument is not sound, but it is informative paying attention to what they consider "evidence" for AGI. A nice instance of a problem that seems peculiar to AI: it tries to define both its target phenomenon and how well it is doing towards it.