Not sure I understand the article. The author specifically chose art from humans and AI that he found difficult to categorize into human or AI art. The fact that people had a 60% success rate suggest that they are a little better in seeing the difference then he was himself?

(What am I missing? This is not like "take 50 random art objects from humans and AI", but take 50 most human like AI, and non-obvious human art from humans)