It’s especially galling given that the current AI du jour, LLMs, don’t do mutch more than reflect their training data back at us. Which means that if they could answer the problem, it would be because people had already answered the question