Last month I wrote a blogpost in which I described how ChatGPT made up academic references after I asked it to suggest resources on Emir Abdelkader. The references seemed deceptively authentic, as if ChatGPT had been intent on misleading me.

Later on, I wanted to ‘confront’ ChatGPT with my accusations, so I asked it to analyze that post. Its answer was: