The author admits that he had a hard time differentiating authenticity of generated text on topics he was not interested in. This probably indicates that he might also have trouble detecting value in those topics even when discussed by a real human. Suggesting that people are not truly reasoning when speaking or do not exhibit general intelligence because they cannot solve assigned math problems was particularly infantile. I think there is just some heavy bias towards formal logic here, devaluing less easily analyzable forms of intelligence.

I doubt the school teachers who failed his mathematics exam would have major issues finding at least some of the problems in the generated texts he gave as examples.

So asking the question whether humans are "artificial" stems from a place of low empathy, I think.