Given this new information about the author's practices I'm inclined to think that everything they say in the above article should be treated with suspicion, especially since they accept that a) they do not have ready access to GPT-3 and b) they edited the "GPT-3" article to some extent. For example I would not be surprised to learn that, unable to use GPT-3 for a significant amount of time, they genereated some of the text of the article and then created the rest by hand, or similar. Not because I don't believe that readers on the internet can mistake GPT-3 as something written by a human, but because the author's ethics sound just... dodgy.