This black spatula case was pretty famous and was all over the internet. Is it possible that the AI is merely detecting something that was already in its training data?
I found many sources on the internet that state that the error in the black spatula paper was discovered by John Schwarcz of McGill University while reviewing a research paper:
