I think this highlights the dichotomy of AI use and how it's shaping everyone's opinion based on their own experience. It's your AI versus mine. You could be OpenAI with unlimited compute and disprove a conjecture or you could be the people referred in the article who are asking claude if a story is written by a human. Opus 4.7 can generate working code faster than I ever could but I still see it as a dumb word calculator bc of the mistakes it makes.