According to it, i fall in the power user category, which is 1 to 5% of its user base. But my calibration is local to me. I have broken it a few times and helped it model things (local to my chatgpt).
Higher the intelligence of the question, higher the intelligence of the answer. Chatgpt that...