I'd go further. Put them in an unlit padded room that has a TV screen so that they have no other sensory input than visual. Only show them a few captioned imag

I'd go further. Put them in an unlit padded room that has a TV screen so that they have no other sensory input than visual. Only show them a few captioned images in repeated succession for brief periods. It's absurd for people who are experts in the field to compare human systems with current state of the art Deep Learning, and it's getting frustrating.